irc.oftc.net #zumastor log beginning Sat Dec 1 00:00:01 PST 2007 2007-12-01 02:58 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-01 04:15 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-01 06:23 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-01 06:24 http://zumastor.org/demo-vmware-ubuntu.html isn't linked yet, but at least it boots now... 2007-12-01 09:56 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-01 10:04 -!- pgquiles_(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-01 11:33 shapor around? 2007-12-01 12:20 -!- pgquiles__(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-01 12:27 -!- pgquiles_(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-01 13:10 -!- pgquiles__(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-01 13:31 -!- pgquiles_(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-01 17:11 -!- SEJeff(~jeff@cpe-76-174-205-207.socal.res.rr.com) has joined #zumastor 2007-12-01 20:42 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-01 21:56 flips, i am now 2007-12-01 23:38 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sun Dec 2 00:00:01 PST 2007 2007-12-02 01:29 shapor, and now? 2007-12-02 02:56 flips, back again 2007-12-02 03:40 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-02 07:24 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-02 07:30 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-02 08:24 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has left #zumastor 2007-12-02 08:25 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-02 13:04 -!- SEJeff(~jeff@cpe-76-174-205-207.socal.res.rr.com) has joined #zumastor 2007-12-02 15:20 flipz: ping 2007-12-02 20:59 -!- SEJeff(~jeff@cpe-76-174-205-207.socal.res.rr.com) has joined #zumastor 2007-12-02 23:42 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Dec 3 00:00:01 PST 2007 2007-12-03 00:39 shapor: what kernel config should I use? -um-uml? -qemu? -full? 2007-12-03 00:40 (for production) 2007-12-03 01:12 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-03 07:59 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-03 08:14 -!- Brad(~Brad@c-24-3-205-220.hsd1.mn.comcast.net) has joined #zumastor 2007-12-03 08:54 -!- maximadelman(~maxim@mail.violintech.net) has joined #zumastor 2007-12-03 09:36 pgquiles, shapor is on the road 2007-12-03 09:36 will be three hours closer to you 2007-12-03 09:37 sejeff, pong 2007-12-03 09:53 flips, ping 2007-12-03 09:58 hi tim_vimm 2007-12-03 09:59 morning.... 2007-12-03 10:00 hi 2007-12-03 10:01 heading into the office about now 2007-12-03 10:01 thinking lkml announcement 2007-12-03 10:01 Maxim is working on your config 2007-12-03 10:01 Subject: [RFC] Snapshots, replication and a look at the future of storage 2007-12-03 10:01 hmm 2007-12-03 10:02 Subject: [RFC] Volume snapshots, replication and a look at the future of storage 2007-12-03 10:02 no, saying volume weakens it 2007-12-03 10:02 people can figure out "volume" from the body text 2007-12-03 10:02 title...? 2007-12-03 10:02 Subject: [RFC] Volume snapshots, replication and a peek at the future of storage 2007-12-03 10:02 subject line for the lkml post 2007-12-03 10:03 it's almost alarming that I'm 5 years into this project and it's still the "future of storage" 2007-12-03 10:04 Volume snapshots, replication, and nfs performance? 2007-12-03 10:04 for the geek crowd, performance = future 2007-12-03 10:05 and nfs doesn't really sum it up 2007-12-03 10:05 the IT crowd thinks NFS =storage and the geek crown thinks nfs = lame 2007-12-03 10:05 let's pander to the geeks then 2007-12-03 10:05 good choice :) 2007-12-03 10:07 in the first sentence I will say the replication is both for local filesystems and nfs 2007-12-03 10:07 netapp only does it for nfs 2007-12-03 10:07 crying shame 2007-12-03 10:07 "future of" sounds like its still on the drawing board 2007-12-03 10:07 glimpse? 2007-12-03 10:07 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-03 10:08 future of hot swap dram 2007-12-03 10:08 it is, it's at least 2 months in the future 2007-12-03 10:08 I can clarify in the text you can buy one today 2007-12-03 10:09 however the backing store driver (push ram to stable disk) is still in the future, slightly 2007-12-03 10:09 can give that a paragraph 2007-12-03 10:10 mention that while there are applications that can work with volatile ram, most want stable backing storage 2007-12-03 10:26 Volume snapshots, replication and emerging storage technology 2007-12-03 10:30 ACTION is alive 2007-12-03 10:36 pgquiles: -full is the most like a distribution kernel, lots of modules, the other are stripped down 2007-12-03 10:37 i thought you built a gutsy kernel with zumastor though, that sounds ideal 2007-12-03 10:38 those configs we have are really just templates that work with our hardware 2007-12-03 12:13 shapor: ok, thanks 2007-12-03 12:13 shapor: I'm building -zumastor and -xenzumastor packages of gutsy's stock kernel (2.6.22-14) 2007-12-03 12:14 it takes an eternity :-( 2007-12-03 12:19 yeah, it takes forever to build distribution kernel with all those modules 2007-12-03 12:34 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-03 12:55 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-03 12:56 -!- maximadelman(~maxim@mail.violintech.net) has left #zumastor 2007-12-03 12:58 shapor: have you tested performance with different I/O schedulers, SLAB allocators, pre-emption models, etc? 2007-12-03 13:19 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-03 13:35 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-03 13:36 pgquiles: no performance testing with different io schedulers unfortunately, although if you do any we'd be interested in the results :) 2007-12-03 13:36 -!- phillips_(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-03 13:37 shapor: :-) 2007-12-03 13:38 we have some work to do optimizing the number of metadata operations 2007-12-03 13:39 so we know the worst case copy out is quite a bit seek-intensive 2007-12-03 13:40 rather than worrying about the ordering of the metadata operations, we intend to remove most of them :) 2007-12-03 14:06 off to bed 2007-12-03 14:15 shapor: tnx for the typo fix 2007-12-03 14:27 -!- Brad_(~Brad@c-24-3-205-220.hsd1.pa.comcast.net) has joined #zumastor 2007-12-03 14:49 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-03 15:28 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-03 17:14 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-03 17:34 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-03 17:57 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-03 21:21 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-03 22:13 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Tue Dec 4 00:00:01 PST 2007 2007-12-04 00:35 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-04 00:50 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-04 02:05 -!- Brad__(~Brad@c-24-3-205-220.hsd1.mn.comcast.net) has joined #zumastor 2007-12-04 03:02 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-04 03:05 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has left #zumastor 2007-12-04 03:36 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-04 05:29 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-04 05:30 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-04 06:59 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-04 07:22 ping 2007-12-04 08:45 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-04 09:47 png 2007-12-04 09:47 pong 2007-12-04 09:48 morning 2007-12-04 09:48 'sup? 2007-12-04 09:48 I understand the NJ server is back up 2007-12-04 09:48 let me know if we have any needs 2007-12-04 09:48 hi tim 2007-12-04 09:49 hi Jaiying 2007-12-04 09:49 yes, the NJ server is back 2007-12-04 09:49 great 2007-12-04 09:51 Somebody was just wishing for something like zumastor: https://www.redhat.com/archives/linux-lvm/2007-December/msg00006.html 2007-12-04 09:51 I wrote him back privately 2007-12-04 10:25 the proposal is somewhere between a mirror and remote replication 2007-12-04 10:26 probably he would be happy with snapshot replication just as we do it 2007-12-04 10:40 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-04 10:40 oh my, I finally got that damn xen+zumastor kernel built a-la debian/ubuntu! 2007-12-04 10:48 hard? 2007-12-04 10:54 flip: yes 2007-12-04 10:54 flips: I added a new custom flavour to the ubuntu/debian scripts and it failed a few times 2007-12-04 10:55 the sad part is it takes forever for the kernel packages to build and usually it failed when it was to link or package the stuff, i. e. after 4-5 hours :-/ 2007-12-04 10:56 I'm uploading the source packages to my Ubuntu PPA, let's hope packages are available in about 10 hours for everybody 2007-12-04 10:56 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-04 10:56 ubuntu-specific and restricted modules are still missing but I'll have them tomorrow, that's easy and fast to build 2007-12-04 11:04 pgquiles, what you did sounds like exactly what we wanted 2007-12-04 11:18 flips: good :-) 2007-12-04 11:20 with a bit of luck, some debian/ubuntu developer will push the -zumastor and -xenzumastor flavours into the official .diff.gz 2007-12-04 11:21 http://ppa.launchpad.net/pgquiles/ubuntu/pool/main/l/linux-source-2.6.22/ for the sources, still needing some cleaning in the debian/abi directory 2007-12-04 11:26 I suppose we should go seek out an interested maintainer somehow 2007-12-04 11:27 which mailing list is right, to offer a package for maintaining? 2007-12-04 11:35 what do you want? a maintainer for your existing packages? 2007-12-04 11:35 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-04 11:36 or do you want to achieve the "maintainer" status in Debian? 2007-12-04 11:36 if the latter, ask in #debian-mentors and the mentors mailing list 2007-12-04 11:36 probalby we want to maintain a basic package on the site, but don't want to do the full job of packaging for debian or ubuntu 2007-12-04 11:36 in the former, no idea. I guess mentors is the right place again. 2007-12-04 11:36 sounds like the right place 2007-12-04 11:37 :-) 2007-12-04 11:37 we do have a debian maintainer on the project by the way, Drake 2007-12-04 11:37 for the -zumastor and -xenzumastor to be officially part of the kernel, the kernel team should approve it 2007-12-04 11:37 if you have a debian maintainer you have 90% of the way already done 2007-12-04 11:37 and actually, drake is the expert, he can probably just find us a willing maintainer 2007-12-04 11:37 right 2007-12-04 11:38 does drake hang around this channel anytime? I have a debian package I'm in need of a sponsor 2007-12-04 11:38 we might actually make that a -ddsnap kernel, because it is not zumastor-specific 2007-12-04 11:38 oh, fine 2007-12-04 11:39 it's just naming 2007-12-04 11:39 there is exactly one place where we touch the kernel and it is not for ddsnap, that is our nfs suspend/resume 2007-12-04 11:39 pgquiles, can you hand us a recipe for repeating your build? 2007-12-04 11:39 without that hack you can't have snapshot rollover on a nfs filesystem exported from a downstream replciation target 2007-12-04 11:40 s/hack/feature/ 2007-12-04 11:40 dkegel: [20:21] http://ppa.launchpad.net/pgquiles/ubuntu/pool/main/l/linux-source-2.6.22/ for the sources, still needing some cleaning in the debian/abi directory 2007-12-04 11:40 :-) 2007-12-04 11:42 flips, don't we also touch the kernel in the bio throttling? 2007-12-04 11:44 dkegel: yes, you do 2007-12-04 11:44 actually, nfs-suspend is only one of the seven patches in the -zumastor flavour 2007-12-04 11:44 pgquiles@pravetz:~/personal/deb/kernel/linux-2.6.22/debian/binary-custom.d/zumastor/patchset$ ls 2007-12-04 11:44 001-nfs-suspend.patch 003-bio-throttle.patch 005-less-throttle-failed.patch 007-remove-fnsd-less-throttle.patch 2007-12-04 11:44 002-AUTO-ddsnap-files.patch 004-ddsnap-base.patch 006-alloc-pages-stackdump.patch 2007-12-04 11:49 I have not tested the kernel yet, btw 2007-12-04 11:52 the whole stuff takes almost 6 hours to build in a xeon dual core with 4GB RAM 2007-12-04 12:10 pgquiles, flips says -zumastor is fine 2007-12-04 12:11 I argue that selling two new names at once (ddsnap and zumastor) is hard, and the more we can keep to just one name, the easier we'll have it 2007-12-04 12:12 dkegel, indeed, we touch the kernel in a few places for stability 2007-12-04 12:12 will your patches be merged into mainstream kernel? 2007-12-04 12:13 ok, I agree, the zumastor name is not only familiar, it's right 2007-12-04 12:14 guys, do you have an opinion on ata-over-ethernet? It looks good and at the same time scary, as it depends on a single company 2007-12-04 12:16 i've used aoe before 2007-12-04 12:16 with hardware from said company :) 2007-12-04 12:18 shapor: what do you think? should I go with iSCSI (well-known, standard minus bugs, etc) or try AoE and pray to Saint Coraid? 2007-12-04 12:18 well when i was using aoe, iscsi really sucked in linux 2007-12-04 12:19 and since then i haven't tried iscsi 2007-12-04 12:19 it was a very attractive way to get lots of disk online 2007-12-04 12:19 cost was the main factor 2007-12-04 12:20 you could buy 250gb ide hdd's and plug them all in cheap 100mbit switches with a gig uplink to a linux box 2007-12-04 12:20 now what they sell are essentially supermicro 15 bay cases 2007-12-04 12:20 with sata controllers 2007-12-04 12:20 pgquiles, last time I tried, the linux AoE server had major performance problems, as in 1/5th of expected throughput 2007-12-04 12:20 and a plan-9 based embedded os on a flash card 2007-12-04 12:21 the hardware aoe boards and the plan 9 one are great aoe targets 2007-12-04 12:21 the linux aoe target sucked more than the iscsi one did at the time 2007-12-04 12:21 it was nothing more than a proof of concept 2007-12-04 12:21 flips: do you mean 1/5th of expected throughtput *linear* or only at the highest speeds? 2007-12-04 12:21 and i dont believe anyone has worked on it since then 2007-12-04 12:21 _linear_ 2007-12-04 12:21 i dont think coraid wants the linux target to perform well 2007-12-04 12:21 then they can't sell supermicro boxes with their little closed os on it 2007-12-04 12:22 major brain damage in the server, I did not track it down as I was busy at the time 2007-12-04 12:22 :) 2007-12-04 12:22 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-04 12:22 I guess I'll go with iSCSI then 2007-12-04 12:22 the code looked fairly decent too 2007-12-04 12:22 mabe it was congestion_wait? 2007-12-04 12:22 pgquiles: yeah "go with iscsi" would be more short answer :) 2007-12-04 12:22 flips: their initiator driver isn't bad either 2007-12-04 12:23 I was going to say "I'll try AoE if I have some spare time" but I know I won't :-) 2007-12-04 12:23 back then when you called coraid tech supper you got the guy who wrote the driver on the phone 2007-12-04 12:23 quite nice 2007-12-04 12:23 i doubt its still that way 2007-12-04 12:23 support* 2007-12-04 12:23 shapor, the company supports the AoE driver, hates and ignores the russion-built AoE server 2007-12-04 12:23 ah right 2007-12-04 12:23 they didnt write that 2007-12-04 12:23 there is a second commercial target (storport) but it's closed source => even scarier :-/ 2007-12-04 12:24 aoe is much better in theory 2007-12-04 12:24 I doubt it would take much to make the linux AoE driver work really well 2007-12-04 12:24 i always that it was silly to put block device storage over ip 2007-12-04 12:24 AoE server I meant 2007-12-04 12:24 aoe is much simpler 2007-12-04 12:24 and being on the same physical segment of the network isn't usually a problem 2007-12-04 12:25 shapor, it amounts to the same thing, iscsi, aoe and nbd all simply export a block device 2007-12-04 12:25 the simplest approach is nbd 2007-12-04 12:25 yeah, but aoe is far simpler than iscsi 2007-12-04 12:25 and coincdentally, the fastest 2007-12-04 12:25 right 2007-12-04 12:25 with an 8-page specification, one could even write its own target and initiator :-) 2007-12-04 12:25 yes, we were pegging the gige port on a 24 port switch 2007-12-04 12:25 I would much rather work with AoE over iscsi, if the server worked properly 2007-12-04 12:26 then upgraded to a switch with 2x gig and pegged that too 2007-12-04 12:26 flips: ¿AoE over iscsi? :-? 2007-12-04 12:26 oh, yeah 2007-12-04 12:26 :P 2007-12-04 12:26 my bad 2007-12-04 12:26 s/over/rather than/ 2007-12-04 12:26 :) 2007-12-04 12:26 I though you meant a second layer :-D 2007-12-04 12:26 I meant AoE vs iscsi, yes 2007-12-04 12:29 -zumastor and -xenzumastor are already building in PPA, they'll be available in +5 hours 2007-12-04 12:37 ACTION <- big smile 2007-12-04 13:00 ACTION goes to sleep 2007-12-04 13:02 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-04 13:03 hi jiayingz 2007-12-04 13:06 ok, coming in 2007-12-04 13:09 hi flips 2007-12-04 13:12 hi jiayingz, do we have new nfs transaction latency numbers from the violin now, without separate metadata? 2007-12-04 13:23 I got some data with separate metadata 2007-12-04 13:25 how did that look, similar to yesterday? 2007-12-04 13:26 is today's with 16k metadata blocks, and 16k chunksize? 2007-12-04 13:27 -!- maximadelman(~maxim@mail.violintech.net) has joined #zumastor 2007-12-04 13:29 yes. but I got a similar problem as yesterday 2007-12-04 13:29 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-04 13:29 may need to reboot the machine again 2007-12-04 13:30 maxim's telephone number might be useful in that case 2007-12-04 13:31 not sure if it is our bug 2007-12-04 13:33 without separate metadata would be a good test 2007-12-04 13:33 obviously we do need to track down the bug 2007-12-04 13:33 but just avoid for now 2007-12-04 13:47 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-04 13:52 somebody needs to push the 2.6.23 rebased patches from ben m into svn 2007-12-04 13:52 I suppose that would be me since I acked them 2007-12-04 13:53 it would be nice to have a system, but the 7 patches or so need about 4 different methods of applying them between them 2007-12-04 13:53 hard to automate 2007-12-04 13:55 I believe we have iSCSI instead of SCSIoE because certain parties wanted to be able to use various funky and often proprietary networking hardware... beyond that, it isn't clear to me why we even have AoE instead of SCSIoE, given how similar the command sets are these days. 2007-12-04 13:57 flips, yes, please do grab those patches. 2007-12-04 14:09 hmm, the mm version of bio.c has changed bio_endio to eliminate byes_done, a suggestion I made while getting flamed by Jens 2007-12-04 14:09 go figure 2007-12-04 14:15 the bio.throttle patch got a little shorter when rebased to -mm 2007-12-04 14:15 trivial things 2007-12-04 14:26 but good ones 2007-12-04 14:32 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-04 14:57 -!- maximadelman(~maxim@mail.violintech.net) has left #zumastor 2007-12-04 14:59 Hey, how beneficial is the DM_DDSNAP_SWAP option? 2007-12-04 15:13 not needed with new 2.6.22.10 kernel 2007-12-04 15:29 jiayingz: all is good? 2007-12-04 15:45 I just asked maxim to reboot the machine again 2007-12-04 15:45 there seemed to be some power failure 2007-12-04 15:47 everything is ok so far after he moved the machine to another circuits 2007-12-04 16:36 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-04 17:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-04 17:48 jaiyingz: just sms'd Maxim 2007-12-04 17:49 don't know if anyone is still there 2007-12-04 17:55 thx, Tim 2007-12-04 18:24 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-04 18:24 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-04 18:24 -!- juuva(juuva@peili.org) has joined #zumastor 2007-12-04 19:49 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-04 21:55 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-04 22:51 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Wed Dec 5 00:00:01 PST 2007 2007-12-05 00:30 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-05 02:49 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has left #zumastor 2007-12-05 06:27 -!- szhocker(~root@bxn46.internetdsl.tpnet.pl) has joined #zumastor 2007-12-05 06:27 hi 2007-12-05 06:27 I have a question 2007-12-05 06:28 what it the max partition size sypportet by zumastor 2007-12-05 07:41 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-05 10:15 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-05 10:15 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-05 11:39 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-05 11:55 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-05 12:20 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-05 12:39 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-05 12:43 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-05 12:44 szhocker: sorry we missed you. In 0.4 it's 2TB due to a bug (http://code.google.com/p/zumastor/issues/detail?id=10). I don't know what the limit is now. What size are you interested in? 2007-12-05 12:44 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-05 12:45 dkegel: is it appropriate to compliment your bumper sticker collection here? 2007-12-05 12:48 :-) 2007-12-05 12:48 Never mix tech and politics :-) 2007-12-05 12:48 All the tree-huggers here wouldn't approve of my "alt.pave.the.earth" bumper sticker :-) 2007-12-05 12:49 :) 2007-12-05 13:10 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-05 13:10 blah, hadoop 2007-12-05 13:14 sorry, wrong channel 2007-12-05 13:33 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-05 13:34 mmm the -xenzumastor kernel generates bad code and it does not boot in ubuntu server 7.10 :-/ 2007-12-05 13:35 I guess those "section mismatch" warnings were meaningful 2007-12-05 13:35 ouch, usually are. 2007-12-05 13:35 BTW, Tim, see http://www.geocities.com/SouthBeach/1380/pave.html 2007-12-05 13:36 have you built the xen one before? 2007-12-05 13:36 without zumastor, I mean? 2007-12-05 13:38 dkegel: yes 2007-12-05 13:38 dkegel: -xen works fine 2007-12-05 13:38 and I'm booting my -zumastor right now 2007-12-05 13:41 yes, -zumastor boots fine 2007-12-05 13:41 finding out why zumastor + xen patches results in bad code is going to be "funny" 2007-12-05 13:43 esp given that each full build of the ubuntu kernel with all its flavours takes about 6 hours :-/ 2007-12-05 13:43 Did you post what the section mismatch was? 2007-12-05 13:44 no 2007-12-05 13:44 the problem is even without patches, there are several section mismatch errors 2007-12-05 13:45 but those generate code which is OK 2007-12-05 13:45 odd 2007-12-05 13:45 yeah 2007-12-05 13:46 http://launchpadlibrarian.net/10736761/buildlog_ubuntu-gutsy-amd64.linux-source-2.6.22_2.6.22-14.46~ppa2_FULLYBUILT.txt.gz 2007-12-05 13:46 full build log 2007-12-05 13:46 "enjoy" :-D 2007-12-05 13:46 Is the irony *really* necessary? ;-) 2007-12-05 13:55 bbiab 2007-12-05 14:12 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-05 14:22 mmm actually, -xen is corrupt too 2007-12-05 14:23 oddly there are no more warnings or section mismatchs in -zumastor than in -xen or -xenzumastor 2007-12-05 14:23 probably it's the .config I'm using for -xen and -xenzumastor, which is different than the one I used for -zumastor 2007-12-05 14:40 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-05 15:10 oh joy, -xenzumastor is working! 2007-12-05 15:10 the magic was? 2007-12-05 15:11 the magic was ubuntu is stupid and grub fails to recognize xen-ified kernels unless xen is already installed 2007-12-05 15:12 which is more or less logical but, please, add a dependency on ubuntu-server-xen for linux-image-xen! 2007-12-05 15:12 hehe 2007-12-05 15:12 bug #147014 btw 2007-12-05 15:13 good call 2007-12-05 15:13 I'll add that dependency to the packages in my PPA 2007-12-05 15:14 who should I kill to have my PPA included in the downloads section? (as "unofficial packages" maybe?) 2007-12-05 15:15 My lawyer tells me that if I suggest someone I'll be guilty as an accessory. ;-) 2007-12-05 15:16 :-D 2007-12-05 15:17 I'll just bug people until someone bites :-) 2007-12-05 15:17 Good call. The sentences for stalking are much lighter. ;-) 2007-12-05 17:23 off to bed 2007-12-05 21:01 PPA? 2007-12-05 21:01 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-05 21:01 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-05 21:03 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-05 22:23 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-05 22:24 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has left #zumastor irc.oftc.net #zumastor log beginning Thu Dec 6 00:00:01 PST 2007 2007-12-06 01:38 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-06 01:50 Going to sleep in today... 2007-12-06 01:50 ACTION zzzzz 2007-12-06 07:03 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 07:47 -!- pgquiles(~pgquiles@205.Red-80-39-172.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 08:09 thanks your adding information about my packages for ubuntu in zumastor.org 2007-12-06 09:18 -!- cbsmith(~cbsmith@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-06 09:24 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 09:47 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-06 10:22 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 10:41 "Zumastor uses about 1MB of RAM per gigabyte of disk space to cache an internal index" 2007-12-06 10:41 wow, this is really inconvenient 2007-12-06 10:42 it means I need 3GB of RAM just for the indices in my servers with 3TB of hard-disk :-/ 2007-12-06 10:51 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-06 10:51 that's to run optimally, otherwise it will be reloading a lot of index from cache 2007-12-06 10:51 from disk 2007-12-06 10:51 and it depends on how heavily churned your main volume is 2007-12-06 10:52 if lightly churned, then much less is needed 2007-12-06 10:52 we should explain the cache issue better 2007-12-06 10:53 anyway, your server most probably does have 3 GB available for caching if you need it 2007-12-06 11:08 it also depends on your chunk size 2007-12-06 11:17 true 2007-12-06 11:17 ah, and the units of measure are wrong, it is 1 MB of index per 1 TB of _snapshot store_ 2007-12-06 11:18 sorry for not picking that up sooner 2007-12-06 11:18 flips: Oh, that's very different. 2007-12-06 11:19 flips: And it makes more sense too. :-) 2007-12-06 11:23 -!- pgquiles__(~pgquiles@205.Red-80-39-172.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 11:25 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 11:25 hey, mitchg, welcome 2007-12-06 11:25 Hello 2007-12-06 11:33 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 11:43 flips ping 2007-12-06 11:43 pong 2007-12-06 11:44 ping pong 2007-12-06 11:44 How about /var/lib/zumastor//filesystem/mountpoint, to go with /var/lib/zumastor//filesystem/options that mount_device already knows about? 2007-12-06 11:45 nice I think 2007-12-06 11:45 better than /mount/ 2007-12-06 11:45 It's already there (in the code, anyway), might as well use it. 2007-12-06 11:45 Yes. 2007-12-06 11:46 -!- pgquiles_(~pgquiles@91.Red-81-44-156.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 11:46 I also plan to create the file in start_volume if it doesn't already exist... 2007-12-06 11:48 shapor: Review that patch! >:( 2007-12-06 11:50 fmayhar sorry been pretty busy 2007-12-06 11:50 No prob. Just impatient. :-) 2007-12-06 11:51 What's special about "unqualified?" 2007-12-06 11:52 On mount_device calls. 2007-12-06 11:53 -!- pgquiles_(~pgquiles@80.Red-81-37-88.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 12:13 -!- pgquiles__(~pgquiles@217.Red-81-39-193.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 12:15 flips: what does "churned" mean? 2007-12-06 12:15 and yes, I planned on 3TB of _snapshot_ store 2007-12-06 12:16 my servers currently have 4GB of RAM 2007-12-06 12:16 if zumastor needs 3GB for the indices... :-/ 2007-12-06 12:17 pgquiles__: Not some much "needs" as "wants". ;-) 2007-12-06 12:18 cbsmith: do you mean it won't start swapping if I don't have 3GB of free RAM available :-? 2007-12-06 12:19 pgquiles__: i think it has a maximum about of ram it will consume, based off a % of system memory iirc 2007-12-06 12:19 after that it will have to go to the disk for snapshot metadata blocks 2007-12-06 12:20 shapor: that sounds pretty acceptable to me, thank you 2007-12-06 12:20 shapor: any plans to fix bug 5 any time soon? 2007-12-06 12:22 flips: ^ ? 2007-12-06 12:23 flips: when do we plan on having an extent-based tree? :) 2007-12-06 12:28 it will be several weeks of somebody hiding and doing nothing but coding 2007-12-06 12:28 followed by stabilization 2007-12-06 12:29 so whenever that fits in the schedule and gets to the top of the list 2007-12-06 12:29 I could start floating detailed design ideas 2007-12-06 12:30 then anybody who feels like taking a run at it could. Beware: don't underestimate the difficulty 2007-12-06 12:31 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-06 12:31 ok, time to hike over to the office 2007-12-06 12:31 good day maze 2007-12-06 12:31 hey 2007-12-06 12:37 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-06 12:42 -!- pgquiles__(~pgquiles@139.Red-81-39-192.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 13:59 -!- pgquiles_(~pgquiles@133.Red-83-49-101.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 14:08 -!- pgquiles__(~pgquiles@24.Red-83-35-113.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 14:19 -!- pgquiles_(~pgquiles@241.Red-83-34-135.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 14:54 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-06 14:54 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 14:54 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 14:54 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 14:54 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 14:54 -!- flipz(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 14:54 -!- flips(~phillips@phunq.net) has joined #zumastor 2007-12-06 14:54 -!- juuva(juuva@peili.org) has joined #zumastor 2007-12-06 14:54 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-06 14:54 -!- pgquiles_(~pgquiles@241.Red-83-34-135.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 14:54 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 14:55 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 18:17 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-06 18:29 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 18:37 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 18:41 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 19:17 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-06 21:21 -!- pgquiles__(~pgquiles@103.Red-217-125-197.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 21:21 -!- flips(~phillips@phunq.net) has joined #zumastor 2007-12-06 21:21 -!- flipz(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 21:21 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 21:21 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 21:21 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 21:21 -!- juuva(juuva@peili.org) has joined #zumastor 2007-12-06 21:21 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-06 21:21 -!- pgquiles_(~pgquiles@241.Red-83-34-135.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-06 21:21 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-06 21:35 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 21:42 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 21:44 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 21:51 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 21:57 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 22:02 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 22:03 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 22:06 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 22:10 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-06 22:13 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 22:19 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 22:30 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 22:41 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 22:45 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-06 22:49 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 23:01 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 23:08 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 23:13 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-06 23:15 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has left #zumastor 2007-12-06 23:16 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 23:23 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 23:30 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 23:37 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-06 23:41 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor irc.oftc.net #zumastor log beginning Fri Dec 7 00:00:01 PST 2007 2007-12-07 01:45 -!- pgquiles_(~pgquiles@132.Red-83-33-144.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 02:02 -!- pgquiles__(~pgquiles@132.Red-83-33-144.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 03:28 -!- MaZe(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-07 03:38 -!- nataliep_(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-07 03:40 -!- juuva(juuva@peili.org) has joined #zumastor 2007-12-07 04:56 -!- pgquiles_(~pgquiles@103.Red-81-36-194.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 05:08 -!- pgquiles__(~pgquiles@201.Red-81-39-193.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 06:50 -!- pgquiles__(~pgquiles@170.Red-81-34-4.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 06:55 -!- pgquiles_(~pgquiles@252.Red-81-35-184.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 07:08 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-07 08:26 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-07 10:01 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-07 10:48 -!- MaZ1(~MaZe@c-67-188-123-1.hsd1.ca.comcast.net) has joined #zumastor 2007-12-07 10:51 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-07 11:59 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-07 12:01 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-07 12:33 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-07 12:36 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-07 12:43 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-07 13:13 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-07 13:16 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-07 14:13 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-07 14:20 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor 2007-12-07 14:33 -!- pgquiles_(~pgquiles@67.Red-83-38-37.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 14:58 -!- Brad(~Brad@c-24-3-205-220.hsd1.mn.comcast.net) has joined #zumastor 2007-12-07 15:12 -!- pgquiles__(~pgquiles@34.Red-83-41-113.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 15:28 -!- pgquiles_(~pgquiles@34.Red-83-41-113.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-07 15:29 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-07 16:42 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sat Dec 8 00:00:01 PST 2007 2007-12-08 01:46 -!- pgquiles__(~pgquiles@18.Red-83-35-242.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-08 02:38 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-08 02:52 -!- pgquiles__(~pgquiles@57.Red-83-41-44.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-08 07:28 -!- pgquiles(~pgquiles@188.Red-217-125-198.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-08 15:08 -!- Brad_(~Brad@c-24-3-205-220.hsd1.mn.comcast.net) has joined #zumastor 2007-12-08 15:27 -!- pgquiles_(~pgquiles@229.Red-217-125-196.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-08 15:55 -!- gabrix(~gabrix@host200-148-dynamic.20-87-r.retail.telecomitalia.it) has joined #zumastor 2007-12-08 18:11 -!- pgquiles_(~pgquiles@229.Red-217-125-196.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-08 19:05 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-08 19:10 ping irc.oftc.net #zumastor log beginning Sun Dec 9 00:00:01 PST 2007 2007-12-09 02:05 -!- pgquiles__(~pgquiles@246.Red-217-125-198.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-09 04:11 -!- pgquiles__(~pgquiles@98.Red-81-35-183.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-09 04:35 -!- pgquiles__(~pgquiles@218.Red-83-41-45.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-09 04:40 -!- pgquiles_(~pgquiles@138.Red-81-39-192.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-09 04:43 -!- pgquiles_(~pgquiles@174.Red-217-125-199.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-09 05:15 -!- pgquiles__(~pgquiles@172.Red-217-125-199.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-09 08:49 -!- Brad__(~Brad@c-24-3-205-220.hsd1.pa.comcast.net) has joined #zumastor 2007-12-09 15:14 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-09 15:29 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-09 22:20 -!- MaZe(~MaZe@216-239-45-4.google.com) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Dec 10 00:00:01 PST 2007 2007-12-10 00:22 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-10 01:35 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-10 01:55 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-10 02:28 flips: +1 to fix bug #6 2007-12-10 02:32 pgquiles, more than 64 snapshots? out of interest, what is your use case? 2007-12-10 02:34 flips: we'd like to have snapshots every few hours for the last month 2007-12-10 02:35 ah, just write the delta to backing store then 2007-12-10 02:35 I don't understand 2007-12-10 02:35 you can generate a delta file for each snapshot if you like 2007-12-10 02:35 you do not have to keep all snapshots online in the snapshot store 2007-12-10 02:39 incidentally, there is an out-of-tree patch for zumastor to be able to mount a snapshot, given a delta file 2007-12-10 02:39 effectively allowing infinite snapshots 2007-12-10 02:39 we need to merge that 2007-12-10 02:40 jonathan's patch, I was reading about that 2007-12-10 02:41 essentially what I want to do is to have a versioned filesystem 2007-12-10 02:41 LVM snapshots aren't there but zumastor looks quite good 2007-12-10 02:42 we're some distance away from versioned filesystem though 2007-12-10 02:42 maybe one day 2007-12-10 02:42 removing the limit on number of snapshots is likely 6-8 months away 2007-12-10 02:43 it will compress the snapshot store metadata at the same time, which is a great reason for doing it 2007-12-10 02:52 actually I'm more worried about bug #5 than bug #6 but since asking is free... :-) 2007-12-10 02:52 the bug seems to exaggerate the issue 2007-12-10 02:54 it conflates snapshot tree size with cache usage 2007-12-10 02:55 "is excessive" should not be in the bug title 2007-12-10 02:57 there was a version-aware Konqueror for KDE 3.x but I cannot find it now :-( 2007-12-10 02:57 when you do find it, I need to know 2007-12-10 02:58 well, 1MB for each 1GB seems excessive to me, a 3TB zumastor'ed volume means 3GB of memory just for zumastor :-/ 2007-12-10 03:01 flips: found it: http://www.sandeepranade.com/html/ComputerScience/time-travelling-file-manager.html 2007-12-10 03:02 adapting the mountable delta patch to this interface would be a sweet hack 2007-12-10 03:03 yes 2007-12-10 03:04 actually, it'd be nice it the time-travelling filemanager worked for Samba- and NFS-mounted volumes 2007-12-10 03:04 that'd require metadata to be readable through Samba/NFS 2007-12-10 03:04 using the .snapshot directory, yes 2007-12-10 03:05 great 2007-12-10 03:05 and a Windows Explorer plugin or something like that to be usable from Windows, too 2007-12-10 03:06 anyway, folks are working on the hookup to the windows "previous versions" api 2007-12-10 03:06 I need to go sleep about now, see you 2007-12-10 03:06 ok 2007-12-10 03:06 bye 2007-12-10 08:40 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-10 10:35 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-10 10:38 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-10 10:41 -!- lidi20(~lidi20@ANantes-257-1-113-73.w90-25.abo.wanadoo.fr) has joined #zumastor 2007-12-10 11:08 -!- MaZe(~MaZe@216-239-45-4.google.com) has left #zumastor 2007-12-10 11:38 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-10 12:24 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-10 12:31 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-10 16:46 -!- willn(~wan@paperboy.ccs.neu.edu) has joined #zumastor 2007-12-10 17:00 FYI, trunk/doc/install.html on googlecode.com still has some stuff that hasn't been sanitized to example.com 2007-12-10 17:02 willn: Can you post references to the list? 2007-12-10 17:09 The mail list? 2007-12-10 17:11 willn: yes. 2007-12-10 17:13 Will do. 2007-12-10 18:47 The howto moved, but there shouldn't be any loose ends left. 2007-12-10 19:30 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-10 22:43 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-10 23:01 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Tue Dec 11 00:00:01 PST 2007 2007-12-11 00:47 cbsmith: I don't know what the guy who reported bug #21 meant or wanted, it could probably be just closed 2007-12-11 00:53 pgquiles: What he meant was... let's say we discover that the filesystem is a total mess, and the last known good state was yesterday. 2007-12-11 00:53 pgquiles: You might, in that case, want to revert to yesterday's snapshot, and have it become the new head. 2007-12-11 00:53 pgquiles: Right now Zumastor can't make it the new head. 2007-12-11 00:54 cbsmith: ah, ok: instead of copying the whole snapshot over head, tell zumastor to change heads 2007-12-11 00:55 pgquiles: Yup, more efficient. 2007-12-11 00:55 pgquiles: particularly if you are doing remote replication somewhere. ;-) 2007-12-11 00:58 then a "revert this file/directory to this snapshot" would probably make sense for bandwidth reasons 2007-12-11 01:01 pgquiles: Yeah, but before you could do that you'd need a way to figure out which blocks belonged to a given file in two different snapshots. 2007-12-11 01:01 Actually, it might even be messier than that. There's the whole issue of reconstructing the parent directories. 2007-12-11 01:03 cbsmith: I know nothing about the interlas of zumastor but I'm sure you are right :-) 2007-12-11 01:03 s/interlas/internals 2007-12-11 01:03 pgquiles: It's a good idea though. Just a lot of work to do it. 2007-12-11 01:05 Maybe someone else has a brilliant idea on how to implement it easily. 2007-12-11 01:28 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-11 04:58 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-11 07:29 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-11 10:56 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-11 10:57 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-11 11:05 cbsmith: http://gentoo-wiki.com/Talk:TIP_Bluetooth_Proximity_Monitor 2007-12-11 11:45 # ./analyze-tree.pl < tree 2007-12-11 11:45 extents would save 1269588 out of 1996607 total entries 2007-12-11 11:45 fmayhar ^ 2007-12-11 12:27 ? 2007-12-11 12:27 shapors script? 2007-12-11 12:28 meaning that many entries have an adjacent neighbour? 2007-12-11 12:28 flips yeah, i posted the script to the list 2007-12-11 12:28 i believe its correct 2007-12-11 12:29 frank was questioning how much benefit there would be in implementing extents in the btree 2007-12-11 12:29 so you answered, in precise terms 2007-12-11 12:29 yeah got me curious, i had a feeling it would be significant 2007-12-11 12:30 "ddsnap dump" :) 2007-12-11 12:30 the size of an entry increases of course 2007-12-11 12:30 yes, I figured 2007-12-11 12:30 that feature earned back the effort of implementing pretty fast 2007-12-11 12:31 after i remembered that the dump command printed to the log file 2007-12-11 12:31 it only took about 15 minutes to write that 2007-12-11 12:31 yay perl 2007-12-11 12:32 yay dump 2007-12-11 12:32 :) 2007-12-11 12:32 oh, yay ascii 2007-12-11 12:32 and yay big disks to hold big dumps 2007-12-11 12:33 it was only about 200MB for the dump 2007-12-11 12:33 but i ran it about 5 times with no output before checking the log 2007-12-11 12:33 even though its for debugging we should document it ;) 2007-12-11 12:33 ok, the savings will not be limited to just encoding neighbours 2007-12-11 12:34 there are also savings in the serial encoding 2007-12-11 12:34 what else? 2007-12-11 12:34 ah 2007-12-11 12:34 of sharemask? 2007-12-11 12:34 and the delta encoding <- another form of serial 2007-12-11 12:34 yes 2007-12-11 12:35 so I guess the final benefit will be a factor of 5 or so 2007-12-11 12:35 maybe just 3 or 4 2007-12-11 12:35 yeah this was pretty simple minded neighbor detection basically 2007-12-11 12:35 because physical addresses still have to be encoded per physical entry 2007-12-11 12:36 and still well over 50%, on a server with mostly small files 2007-12-11 12:36 its encouraging 2007-12-11 12:36 however, those addresses are compressable if they are near each other 2007-12-11 12:36 which they ought to be in an optimized system 2007-12-11 12:36 there will be a slight cpu trade off 2007-12-11 12:36 to encode in any other way 2007-12-11 12:36 even to support extents 2007-12-11 12:37 depends on the encoding, yes 2007-12-11 12:37 we are not going to gzip the leaves 2007-12-11 12:37 the current implementation is quite efficient cpu wise 2007-12-11 12:37 though essentially, we are doing something similar 2007-12-11 12:37 yeah run length encoding or something 2007-12-11 12:37 the new implementation will not be much more costly 2007-12-11 12:37 a few extra adds in lookup 2007-12-11 12:38 probably even lost in the noise of memory access on modern cpus 2007-12-11 12:38 actually, if each new entry is added at the end of a serial blob as I plan, that will be more cpu efficient for add than the current scheme 2007-12-11 12:38 hm 2007-12-11 12:39 considerably more, because memmoves get prertty expensive 2007-12-11 12:39 so if lookup is only a little slower, overall might be a win 2007-12-11 12:39 ah right.. the memmove 2007-12-11 12:39 that will be nice to lose 2007-12-11 12:40 minor blemish 2007-12-11 12:41 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-11 12:41 http://pages.plotinka.ru/~cyciron/chinauser.jpg 2007-12-11 12:42 shapor: Also in email. 2007-12-11 13:41 svn: Commit failed (details follow): 2007-12-11 13:41 svn: PROPFIND request failed on '/svn/trunk/kernel/ddsnap-patches/2.6.22.10' 2007-12-11 13:41 svn: PROPFIND of '/svn/trunk/kernel/ddsnap-patches/2.6.22.10': 502 Bad Gateway (https://zumastor.googlecode.com) 2007-12-11 13:45 flips: awesome 2007-12-11 13:46 retry? 2007-12-11 14:01 flips: that happens all the time 2007-12-11 14:01 retry should fix it 2007-12-11 14:02 i get ~ one of those a day 2007-12-11 14:02 from my updater cronjob which runs every 5 minutes 2007-12-11 14:06 why doesn't it learn to retry itself? 2007-12-11 14:24 I heard that's on the list of things to do 2007-12-11 14:36 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-11 14:43 I guess I'll look at Issue 22: Allow tuning snapshot store cache size 2007-12-11 16:03 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-11 17:09 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-11 23:29 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-11 23:34 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor irc.oftc.net #zumastor log beginning Wed Dec 12 00:00:01 PST 2007 2007-12-12 00:34 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-12 00:48 -!- natalie(~nataliep@66-102-14-1.google.com) has joined #zumastor 2007-12-12 02:23 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-12 08:09 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-12 08:50 http://lkml.org/lkml/2007/12/12/237 2007-12-12 08:51 violin benchmarks 2007-12-12 08:51 awe-some 2007-12-12 08:51 good morning :) 2007-12-12 08:51 good evening 2007-12-12 08:51 I gotta get me summa dat sleep now 2007-12-12 08:51 I figured that 2007-12-12 08:52 hit me when you wake, Rip Van Winkle 2007-12-12 08:52 I didn't actually stay up all night, I tossed and turned a few hours first before I realized that I wasn't going to get any sleep until the post was up 2007-12-12 08:52 I hate that feeling 2007-12-12 08:52 sorta sleeping, but mostly not 2007-12-12 08:53 I'll have a cuppa tea first and see if there are any early replies 2007-12-12 09:07 -!- flipz(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 09:07 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 09:07 special extra thanks to the zumastor team 2007-12-12 09:08 the engineering team is having a bit of a party right now 2007-12-12 09:08 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 09:21 heh 2007-12-12 09:27 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 09:27 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 09:29 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 10:48 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-12 10:53 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 12:47 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-12 12:57 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-12 13:25 -!- erwan__taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-12 13:33 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-12 13:58 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-12 14:07 aha. I just had to add SUBARCH=i386 when building UML, I bet. 2007-12-12 14:10 dank: Oh, you were building on a 64-bit kernel? 2007-12-12 14:10 dank: Sorry, if I'd realized, I could have told you that. You may also have to use an older kernel, as UML has some issues with the SUBARCH with 2.6.22 IIRC (maybe it is just 2.6.23). 2007-12-12 14:11 yes, 32 bit userspace, 64 bit kernel. 2007-12-12 14:11 dank: Yup, need subarch for that. 2007-12-12 14:11 dank: On the co-op we're actually still using 2.6.18 for our usermode kernels. 2007-12-12 14:12 Bleah. So I guess I'll add a line to build_uml.sh to check uname -m and add the SUBARCH if this works. The kernel did build ok, now to see if it boots. 2007-12-12 14:12 dank: You could always build a 64-bit UML kernel. That should work now. 2007-12-12 14:13 But I don't have a 64 bit userland. 2007-12-12 14:13 Outside uml, I mean. 2007-12-12 14:13 Oh right... silly me. 2007-12-12 14:14 I think you only need a 64-bit libc, but that is probably a tall order. 2007-12-12 14:15 dank: So then, I'd say the build script *shouldn't* automatically do "uname -m" to make the call to use SUBARCH, because there are lots of people for whom the problem is the opposite (64-bit runtime, no 32-bit runtime), particularly with say... NFS servers. 2007-12-12 14:16 Oh, I would only do it if userland was 32 bits. 2007-12-12 14:16 So first you should detect that. 2007-12-12 14:16 Or does the build already do that? 2007-12-12 14:23 Probably not. 2007-12-12 14:24 Create origin device...FAIL 2007-12-12 14:24 darn. 2007-12-12 14:38 what's the effect of the many seeks ddsnap performs on the life of the hard-disks? have you studied that? 2007-12-12 14:40 in other words: how fast will ddsnap kill my SATA hard disks? 2007-12-12 14:44 pgquiles: No studies, but I'd imagine it'd work them harder. 2007-12-12 14:45 pgquiles: If you are really worried, use a RAID controller with battery backed RAM. :-) 2007-12-12 14:53 cbsmith: I'm using that :-) 2007-12-12 14:53 Smart Array P400, 512MB BBWC 2007-12-12 14:53 HP 2007-12-12 14:57 pgquiles: That should significantly reduce the number of seeks that you see on the actual disk. Most of those extra seeks involve updating tiny meta data that would get cached in RAM. 2007-12-12 14:59 it'd be nice to know how much RAM should you use depending on the volume size, ie if my zumastored volume is 1TB, do I need 512MB of cache? or is 256 enough? maybe more? maybe less? 2007-12-12 15:01 pgquiles: We're working on the memory issues now (and we expect them to improve as a consequence), but in short it is a function of the combined size of the volume and the snapshot store. 2007-12-12 17:01 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-12 18:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-12 18:40 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-12 20:56 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-12 21:39 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-12 22:27 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Thu Dec 13 00:00:01 PST 2007 2007-12-13 01:15 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-13 07:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-13 10:41 -!- cbsmith(~xman@69-238-60-140.cbean018.lsan03.wayport.net) has joined #zumastor 2007-12-13 11:06 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-13 13:21 http://reubuntu.blogspot.com/ 2007-12-13 13:33 -!- cbsmith(~xman@69-238-60-140.cbean018.lsan03.wayport.net) has joined #zumastor 2007-12-13 14:41 http://www.engadget.com/2007/12/13/battleship-mtron-the-absurdly-fast-ssd-raid-array/ 2007-12-13 14:51 shapor: waow 2007-12-13 14:52 ACTION would love testing some SSDs 2007-12-13 14:55 they're expensive though 2007-12-13 14:55 another article said $1500 per drive 2007-12-13 14:55 depends on the size 2007-12-13 14:56 ACTION saw some 1.6TB will be released in Q1 2008 2007-12-13 14:56 65K$ 2007-12-13 14:56 or 4K$ for 128GB 2007-12-13 14:56 32G for 400$ 2007-12-13 14:56 128G for 3200$ 2007-12-13 14:57 yep 2007-12-13 14:58 missed you said that. 2007-12-13 14:58 woops 2007-12-13 14:58 :p 2007-12-13 14:58 nice to see the same prices 2007-12-13 14:59 I nearly got a SSD for my laptop, but I couldn't justify the cost. 2007-12-13 14:59 willn: which drives were those prices for? 2007-12-13 15:01 2.5in sata 2007-12-13 15:01 willn: I mean shapor was looking for the brand :p 2007-12-13 15:02 aye. http://www.newegg.com/Product/ProductList.aspx?Submit=ENE&N=2003240636+1421530856&Configurator=&Subcategory=636&description=&Ntk=&srchInDesc= 2007-12-13 15:03 in 2 years, this disk will be affordable 2007-12-13 15:03 it's too early for buying them 2007-12-13 15:03 Buy some of those ram cards, their not terribly expensive 2007-12-13 15:03 and if you can buy ram in any sort of bulk, the storage isnt bad 2007-12-13 15:04 http://techreport.com/articles.x/9312 2007-12-13 15:08 -!- cbsmith(~xman@69-238-60-138.cbean018.lsan03.wayport.net) has joined #zumastor 2007-12-13 16:13 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-13 18:56 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Fri Dec 14 00:00:01 PST 2007 2007-12-14 00:46 shapor, echo "is unnecessary" should be "necessary" and why do we need that line of chatter at all? 2007-12-14 00:51 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-14 01:25 -!- pgquiles(~pgquiles@23.Red-81-36-120.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 02:15 -!- pgquiles_(~pgquiles@69.Red-81-44-156.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 03:51 -!- pgquiles_(~pgquiles@233.Red-81-35-182.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 04:29 -!- pgquiles_(~pgquiles@233.Red-81-35-182.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 05:12 -!- pgquiles_(~pgquiles@190.Red-81-35-182.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 06:59 -!- pgquiles_(~pgquiles@100.Red-217-125-197.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 07:56 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-14 08:45 we dont need it 2007-12-14 09:11 -!- pgquiles__(~pgquiles@100.Red-217-125-197.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 09:57 got a mention on www.ziobudda.net 2007-12-14 09:58 http://www.ziobudda.net/zumastor_nuova_kernel_patch_per_ubuntu_packages 2007-12-14 10:23 -!- pgquiles_(~pgquiles@201.Red-83-35-113.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 10:24 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has left #zumastor 2007-12-14 10:25 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-14 10:34 -!- pgquiles__(~pgquiles@201.Red-83-35-113.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 10:40 -!- pgquiles__(~pgquiles@18.Red-83-38-37.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 10:47 -!- pgquiles__(~pgquiles@5.Red-81-33-102.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 10:52 -!- pgquiles_(~pgquiles@204.Red-81-36-120.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 10:57 -!- pgquiles__(~pgquiles@94.Red-81-35-182.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 11:04 -!- pgquiles_(~pgquiles@137.Red-217-125-197.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 11:09 -!- pgquiles__(~pgquiles@137.Red-217-125-197.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 11:32 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-14 11:39 -!- pgquiles__(~pgquiles@8.Red-217-125-199.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 11:56 -!- pgquiles_(~pgquiles@51.Red-83-33-145.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 12:02 -!- pgquiles__(~pgquiles@63.Red-81-36-253.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 12:08 -!- pgquiles_(~pgquiles@154.Red-80-39-172.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 12:13 -!- pgquiles_(~pgquiles@115.Red-81-33-103.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 12:20 -!- pgquiles__(~pgquiles@141.Red-83-35-242.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 12:25 -!- pgquiles_(~pgquiles@185.Red-81-35-184.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 12:28 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-14 12:58 reading peterz's latest network swap antideadlock patch set 2007-12-14 12:58 which we carried in our tree for a few months as everybody recalls? 2007-12-14 12:58 fixes a problem that we no longer encounter 2007-12-14 12:58 but in theory could 2007-12-14 12:59 namely network receive memory deadlock on the block write path 2007-12-14 13:09 network receive? 2007-12-14 13:09 applies to unix domain sockets as well? 2007-12-14 13:10 everything that can allocate an skb on receive 2007-12-14 13:11 I think unix domain sockets do, that is from memory of a couple years ago, so I am not absolutely sure 2007-12-14 13:12 ah, it doesn't have to be an skb, any memory allocation on receive can deadlock 2007-12-14 13:12 so yes, unix domain sockets too 2007-12-14 13:13 the dealock is _hard_ to trigger though 2007-12-14 13:13 peterz triggers it by throwing tons of unrelated network traffic at a network socket, after pulling and replugging the network cable in the middle of heavy swapping 2007-12-14 13:14 the "tons of unrelated traffic" part is a tad artificial 2007-12-14 13:14 not for networking 2007-12-14 13:14 and even then, it is not clear to me that deadlock can always occur 2007-12-14 13:14 or will ever occur 2007-12-14 13:14 I do not quite see the scenario 2007-12-14 13:15 I see it more as a potential performance problem than a deadlock 2007-12-14 13:15 peterz has not tried this with bio throttling on the writeout paths 2007-12-14 13:15 which seems like a big omission to me 2007-12-14 14:02 -!- pgquiles(~pgquiles@185.Red-81-35-184.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 14:19 -!- pgquiles_(~pgquiles@185.Red-83-35-243.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 14:27 -!- pgquiles__(~pgquiles@22.Red-83-34-135.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-14 15:08 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-14 15:10 trek to the office time 2007-12-14 15:11 wear your good shoes ;-) 2007-12-14 15:11 the jet assist ones? 2007-12-14 15:11 scramjet 2007-12-14 15:52 the scramjet shoes worked great 2007-12-14 17:37 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-14 17:37 ping 2007-12-14 17:46 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-14 17:48 ping 2007-12-14 18:04 http://castle.pricewatch.com/s/search.asp?s=gigabyte+i-ram&group1=1&sci=33&c=System+Memory&i=&srt=t&mi=0&m=&view= + http://castle.pricewatch.com/s/search.asp?s=1gb+dimm&group1=1&sci=33&c=System+Memory&i=&srt=t&mi=0&m=&view= = 4GB seekless metadata device for under $200 2007-12-14 18:09 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-14 19:07 jiayingz, there? 2007-12-14 19:12 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-14 20:02 shapor: I watched an eBay auction where a guy took a 16GB one of those and put it in a case with quad SATA interface to it. 2007-12-14 20:18 Im interested in playing with one of those 2007-12-14 20:20 Too bad they don't offer one in a 3.5 drive form factor, thn you could plug it into a pre-existing backplane 2007-12-14 21:36 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-14 21:44 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sat Dec 15 00:00:01 PST 2007 2007-12-15 02:49 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-15 19:55 cbsmith: hrm, I didn't know they had ones that supported 16GB irc.oftc.net #zumastor log beginning Sun Dec 16 00:00:01 PST 2007 2007-12-16 19:20 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-16 22:10 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-16 22:32 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Dec 17 00:00:01 PST 2007 2007-12-17 00:09 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-17 02:17 -!- erwan_taf(~erwan@217.128.11.69) has joined #zumastor 2007-12-17 04:01 -!- rasecs(~Cesar@201-67-174-39.fnsce704.dsl.brasiltelecom.net.br) has joined #zumastor 2007-12-17 07:13 -!- erwan_taf(~erwan@217.128.11.69) has joined #zumastor 2007-12-17 07:20 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-17 10:32 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-17 11:15 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-17 12:37 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-17 12:38 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-17 12:39 -!- xman(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-17 12:57 hello everybody 2007-12-17 12:58 ACTION waves 2007-12-17 13:37 hm no one bit and said "hello dr nick" 2007-12-17 18:11 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-17 18:41 welcome natalie 2007-12-17 19:23 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-17 19:57 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-17 20:21 -!- Greshnik(~Greshnik@85-013-fixed.vntc.ru) has joined #zumastor 2007-12-17 21:14 -!- erwan_taf(~erwan@217.128.11.69) has joined #zumastor 2007-12-17 22:09 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-17 23:37 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Tue Dec 18 00:00:01 PST 2007 2007-12-18 00:55 -!- erwan_taf(~erwan@217.128.11.69) has joined #zumastor 2007-12-18 02:10 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-18 02:10 Hi pgquiles... 2007-12-18 02:11 hi dank 2007-12-18 02:12 So, about RAM usage: currently, we always just use 128MB for our snapshot index cache. We're adding an option to let you increase this if you find performance suffers. 2007-12-18 02:12 In other words, no, we don't currently allocate 1MB per GB. 2007-12-18 02:12 in the cache (though we do in the on-disk index). 2007-12-18 02:18 awesome! 2007-12-18 02:19 I have been testing other stuff (virtualization: xen vs qemu vs kvm vs ...), making a bit of time while zumastor stabilizes its interface 2007-12-18 02:20 btw, I've just realized I could try a 9TB volume, as my servers are not in production yet 2007-12-18 02:20 12 x 750GB 2007-12-18 02:21 about the benchmarks, tests, etc 2007-12-18 02:22 do they output something I could easily grasp? 2007-12-18 02:25 -!- erwan_taf(~erwan@visage.seanodes.com) has joined #zumastor 2007-12-18 02:44 Yes, fsstress outputs "operations per second". See http://marc.info/?l=linux-kernel&m=119747818323504&w=2 for benchmark results on two different disks (one ramdisk, one single spindle). 2007-12-18 02:45 golly, 9TB testing would be great. 2007-12-18 02:45 Maybe you could try that when we release 0.5 (hopefully soon)? 2007-12-18 02:46 We're quite close to 0.5 already, so you could just test the trunk. 2007-12-18 02:47 http://zumastor.org/graphs/fstress.sata.jpg is a graph of latency vs. NFS ops per second on a single spindle Zumastor NFS server. 2007-12-18 02:47 dank: sure 2007-12-18 02:48 The graph shows both with and without Zumastor, so you can compare the overhead. 2007-12-18 02:48 dank: actually I'm building ubuntu packages for trunk almost daily here, sometimes I even remember to upload it to my PPA 2007-12-18 02:48 Did you get over the boot problems you had earlier? 2007-12-18 02:49 the graph with the violin machine was impressive, btw 2007-12-18 02:49 yeah. That's some hardware. 2007-12-18 02:49 dank: yes, it was a missing dependency 2007-12-18 02:49 Maybe it's time for us to take your packaging and do those builds in our autobuilder? 2007-12-18 02:49 it's only I had not installed xen yet, therefore grub was not able to recognize the xenified kernel 2007-12-18 02:50 Or at least include your packaging stuff in our tree? 2007-12-18 02:50 that would be great 2007-12-18 02:51 How do you want to do it, should you send a patch to the mailing list? 2007-12-18 02:52 mmm 2007-12-18 02:52 it's not a single patch, if you want to provide all the packages ubuntu provides 2007-12-18 02:53 How about putting your stuff in an ubuntu/ subdirectory, like we have the gentoo/ stuff? 2007-12-18 02:54 the problem with the current .deb which zumastor builds is it won't work with several packages: linux-backports-modules, linux-meta, linux-restricted-modules and linux-ubuntu-modules 2007-12-18 02:54 sweet 2007-12-18 02:54 (Maybe in ubuntu/hardy/ ?) 2007-12-18 02:55 currently I'm building for gutsy 2007-12-18 02:56 do you have a pbuilder environment or should I upload binaries? 2007-12-18 02:56 OK, ubuntu/gutsy, then... 2007-12-18 02:56 I don't know, have to ask Drake. My guess is we don't have one but could set one up. 2007-12-18 02:57 It might be nice if you could upload binaries for 0.5 if we can't get it set up in time. 2007-12-18 02:57 I don't really need one, actually 2007-12-18 02:57 I can build for lpia, i386 and amd64 on my ubuntu PPA, then upload to google code 2007-12-18 02:58 but I have no access for more exotic architectures 2007-12-18 02:59 to tell the truth, I have some sparc, ultrasparc and powerpc machines at home but they are too old to build the kernel in a reasonable amount of time :-) 2007-12-18 02:59 We'll probably declare that trunk is a release candidate sometime soon and/or branch, at which point packages from you would be nice. The only architectures we need are the normal Ubuntu 32 and 64 bit Intel, but if anyone asks for the others, then sure. 2007-12-18 03:00 I have an Ultrasparc 10 running Ubuntu that my intern uses as a monitor stand :-) 2007-12-18 03:00 :-D 2007-12-18 03:04 We hadn't been using Google Code for downloads because we hit its size cap, so Shapor does uploads to zumastor.org for us. 2007-12-18 03:04 my machines belong to my vintage computers collection :-) 2007-12-18 03:04 I have an original IBM PC monochrome monitor in the garage. Probably doesn't work anymore... or would explode if I plugged it in... 2007-12-18 03:05 I have a working one of those :-) 2007-12-18 03:05 and next week my high school friend with whom I used to hack on TRS-80 is going to show me one he picked up recently. I think it was used as a prop, but it'll still bring back memories. 2007-12-18 03:06 FOR I=1 TO 10:PRINT I:NEXT 2007-12-18 03:07 OK, bedtime for bonzo here. I'll see if I can interest Drake into building your packages here. Ciao! 2007-12-18 03:07 bye! 2007-12-18 07:34 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-18 07:46 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-18 09:35 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-18 09:37 -!- kconstan(~kconstan@wdas-1.disneyanimation.com) has joined #zumastor 2007-12-18 10:09 hello 2007-12-18 10:10 what's the zumastor team's opinion (if any) of Sun's T2 chip 2007-12-18 10:11 kconstan: howdy 2007-12-18 10:15 hey 2007-12-18 10:15 multi-connecting 2007-12-18 10:15 reading your email now 2007-12-18 10:16 does that make us multi-threaded? 2007-12-18 10:20 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-18 10:44 hi tim_vimm 2007-12-18 10:45 mornin' 2007-12-18 10:47 which email? 2007-12-18 10:47 not yours, kconstan 2007-12-18 10:47 newest member of the zumastor irc chat 2007-12-18 10:53 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-18 11:01 -!- pgquiles__(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-18 11:09 ok, time to finish up revert 2007-12-18 11:26 funny you mention that 2007-12-18 11:27 we were just talking about how wrong it is that we now mount snapshots in the .snapshot directory on the origin 2007-12-18 11:27 and how that makes revert painful for the admin 2007-12-18 11:27 if the origin is corrupted, and you can't look at your snapshots to see which one you want to revert to, thats a problem 2007-12-18 11:33 how could the origin get corrupted like that? 2007-12-18 11:33 you mean, filesystem damage? 2007-12-18 11:34 I have always hated the ~/.snapshot idea as you know, the only thing it has going for it is, it avoids an extra automount entry 2007-12-18 11:35 I think I proved to myself that there is no other way, but if somebody comes up with a workable alternative that would be great 2007-12-18 11:38 yes, filesystem damage 2007-12-18 11:39 i think it will just make life so much more painful, i dont like it at all 2007-12-18 11:39 not to mention all our snapshots will have a .snapshot directory too then 2007-12-18 11:40 the thing is, implementing .snapshot only in the top level isn't going to make people to .snapshot happy anyway 2007-12-18 11:40 they are used to it working in every directory 2007-12-18 11:41 if there is filesystem damage, surely an admin ought to be involved 2007-12-18 11:42 the importance of user being able to get old files back without help in that case is greatly dminished 2007-12-18 11:42 agreed, but the admin's life is going to be much more painful 2007-12-18 11:42 if he has to manually unmount all the snapshots 2007-12-18 11:42 and remount them 2007-12-18 11:42 somewhere else 2007-12-18 11:42 to find which one is good 2007-12-18 11:42 when users are breathing down his neck 2007-12-18 11:42 because the server is down 2007-12-18 11:42 true that 2007-12-18 11:43 even with automagical ~/.snapshot, snapshots have .snapshot in them 2007-12-18 11:43 unless it is *really* automagical 2007-12-18 11:44 just how automagical would that be? 2007-12-18 11:44 not just the readdir vs lookup hack? 2007-12-18 11:45 i dont like how zumastor now messes with the origin 2007-12-18 11:45 anyway, are you saying that snapshots should not be mounted directly on ~/.snapshot? 2007-12-18 11:45 ~ meaning the origin root correct? 2007-12-18 11:46 of each user's home, yes 2007-12-18 11:47 ok, we can't just have one permanent symlink in ~/.snapshot because symlinks are resolved on the client 2007-12-18 11:47 we can have a bind mount there 2007-12-18 11:47 this is nice because bind mounts don't change the filesystem 2007-12-18 11:47 well you do have to create the mountpoint 2007-12-18 11:47 so ~/.snapshot links off to a directory in /var/run/zumastor, which holds the actual snapshot mounts 2007-12-18 11:47 but thats not *too* horrible 2007-12-18 11:48 right, creating the mountpoint would be part of the initial setup 2007-12-18 11:48 in fact, if the mountpoint does not exist we should not create it 2007-12-18 11:48 can't we just use ddlink or something to put some hooks in the vfs 2007-12-18 11:48 that is an opt out for that whole funky mechanism 2007-12-18 11:48 "just" 2007-12-18 11:48 heh 2007-12-18 11:48 yeah, its just a computer 2007-12-18 11:49 one day I will get my hacking boots on and go wade into that 2007-12-18 11:49 there be dragons 2007-12-18 11:49 need some quiet time to get it right, and then it will still only be degrees of it sucks 2007-12-18 11:49 i dont think its easy but this is all ugly 2007-12-18 11:49 and annoying to admins, i think 2007-12-18 11:50 well, the bind mount ~/.snapshot to a directory /var/run/zumastor/snapshots// 2007-12-18 11:50 that was my original idea, which fell out of cache about a year ago 2007-12-18 11:50 i think that sucks 2007-12-18 11:51 thousands of mounts 2007-12-18 11:51 because? 2007-12-18 11:51 not a problem unless that actually hurts performance 2007-12-18 11:51 so amend that to, try that many mounts 2007-12-18 11:51 makes the output of df useless 2007-12-18 11:51 ah, substantive argument 2007-12-18 11:51 grep is your friend 2007-12-18 11:52 so not a stopper 2007-12-18 11:52 whereas having no ~/.snapshot is a stopper 2007-12-18 11:52 what we need is some kind of "mount on link traversal" 2007-12-18 11:53 a callback from the filesystem when it follows the link 2007-12-18 11:53 messy thing 2007-12-18 11:53 right now we have to go the lots of mounts route 2007-12-18 11:53 so jiaying just tried something using a symlink 2007-12-18 11:53 symlinks don't work 2007-12-18 11:53 they do 2007-12-18 11:53 the nfs server follows them 2007-12-18 11:53 if the target is also exported 2007-12-18 11:53 no, they are resolved on the client 2007-12-18 11:54 right 2007-12-18 11:54 needs more exports 2007-12-18 11:54 more exports makes sense 2007-12-18 11:54 more mounts does not 2007-12-18 11:54 well that is not so bad, if no extra automount is needed 2007-12-18 11:54 yeah :) 2007-12-18 11:54 and? 2007-12-18 11:54 it just works 2007-12-18 11:54 the nfs server is smart enough 2007-12-18 11:54 no automount involved 2007-12-18 11:54 ok, then fine 2007-12-18 11:54 we have a winner 2007-12-18 11:55 no the details... 2007-12-18 11:55 now I mean 2007-12-18 11:55 yeah 2007-12-18 11:55 ok, ~/.snapshot is a symlink to what? 2007-12-18 11:55 we're not even talking about ~/.snapshot yet 2007-12-18 11:55 just /volroot/.snapshot 2007-12-18 11:56 so let's talk about ~/.snapshot 2007-12-18 11:56 ok 2007-12-18 11:57 ~/.snapshot -> /var/run/zumastor/snapshot/ 2007-12-18 11:57 and /var/run/zumastor/snapshot// has symlinks? 2007-12-18 11:57 err 2007-12-18 11:57 thats a lot of symlinks to keep up to date 2007-12-18 11:57 and /var/run/zumastor/snapshot// has bind mounts? 2007-12-18 11:58 no bind mount, i think we can do it all with symlinks 2007-12-18 11:58 actually we dont need to update the symlinks every time 2007-12-18 11:58 if we use a different naming convention, yes 2007-12-18 11:58 otherwise with the hourly.0 etc convention, it has to rotate 2007-12-18 11:58 /var/run/zumastor/snapshot//.hourly.0 -> ../hourly.0/ 2007-12-18 11:58 well 2007-12-18 11:59 if we can just rotate the snapshot mounts, not the symlinks, we win 2007-12-18 11:59 we dont even need to do that 2007-12-18 11:59 we can keep mounting the snapshots on yyy-mm-dd-hh:mm:ss 2007-12-18 11:59 and just have top level symlinks which point to the right ones 2007-12-18 12:00 and link to those top level symlinks from the users 2007-12-18 12:00 we can see that as a feature 2007-12-18 12:00 user can ls the symlinks and see the snapshot dates 2007-12-18 12:00 yeah 2007-12-18 12:01 I don't get ../hourly.0/ 2007-12-18 12:01 that seems spooky 2007-12-18 12:01 why 2007-12-18 12:01 goes up 2007-12-18 12:01 thats fine 2007-12-18 12:01 unnatural 2007-12-18 12:01 if man were meant to fly, he would have been born with wings 2007-12-18 12:01 you dont have any? 2007-12-18 12:02 vestigial ones 2007-12-18 12:03 how about this: /var/run/zumastor/mount/volume/origin (origin mountpoint) 2007-12-18 12:03 /var/run/zumastor/snapshot///.hourly.0 -> /var/run/zumastor/mount//hourly.0/ 2007-12-18 12:03 does not cycle 2007-12-18 12:03 yeah thats fine too 2007-12-18 12:04 why the dot on .hourly.0? 2007-12-18 12:04 typo 2007-12-18 12:04 /var/run/zumastor/snapshot///hourly.0 -> /var/run/zumastor/mount//hourly.0/ 2007-12-18 12:04 ok 2007-12-18 12:04 will post that golden line to the list 2007-12-18 12:04 see any flaw? 2007-12-18 12:04 yes 2007-12-18 12:05 i think this is better: 2007-12-18 12:05 /var/run/zumastor/mount/volume/origin (origin mountpoint) 2007-12-18 12:05 /var/run/zumastor/mount/volume/snapshots/ 2007-12-18 12:06 which contains snapshots mounted on yyyy-mm-dd-hh:mm:ss 2007-12-18 12:06 and symlinks from hourly.0, etc 2007-12-18 12:06 then /var/run/zumastor/mount/volume/user.snapshots 2007-12-18 12:06 how about mocking up a directory that looks the way you want it, and post the tree listing to the list? 2007-12-18 12:06 which contains /hourly.0 symlinks 2007-12-18 12:06 ok that will be easier 2007-12-18 12:07 it's hard to explain via the peephole method 2007-12-18 12:07 yeah 2007-12-18 12:07 ok, revert didn't get done again 2007-12-18 12:07 but this is important too 2007-12-18 12:07 see you for lunch 2007-12-18 13:10 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-18 14:16 hi kconstan, there? 2007-12-18 14:18 here 2007-12-18 14:52 flips: You might want to check out http://www.fusionio.com/ 2007-12-18 14:52 flips: Kind of violin-- 2007-12-18 14:52 kconstan, reping 2007-12-18 14:52 cbsmith, yes I know about that 2007-12-18 14:53 plug-in... good luck failing it over 2007-12-18 14:53 flipz: Ah, I didn't. Thoughts on how it compares to violin? 2007-12-18 14:53 I mean, it is possible, just a small matter of coding 2007-12-18 14:53 Seems slower, but perhaps cheaper? 2007-12-18 14:53 also not hotpluggable, raided... not 500 GB 2007-12-18 14:54 it is flash, so slower, not as highly parallizable, so not possible to ever catch up with a 2U box 2007-12-18 14:54 it has its place 2007-12-18 14:54 violin aims at the same $30/GB price point I think, when they bring out their flash box 2007-12-18 19:15 flipz: seen http://open-fcoe.org ? 2007-12-18 19:16 hehe, I was wondering how long before you guys started chatting about that. 2007-12-18 19:17 I'm not clear on why this provides as much of a benefit as folks seem to think. 2007-12-18 19:17 similar to aoe 2007-12-18 19:17 I guess it might somehow be lower overhead than iSCSI. 2007-12-18 19:17 except perhaps the linux target acutally works :) 2007-12-18 19:17 lol 2007-12-18 19:18 Yeah, but a lot of the benefits of FC are in the way the FC layer works... and much of that is lost if it runs on top of ethernet. 2007-12-18 19:18 howso? 2007-12-18 19:18 its just scsi-oe essentially 2007-12-18 19:20 yes, although perhaps slightly more overhead. I'm not entirely clear on the real world differences between AoE and what SCSIoE would be like in the first place. 2007-12-18 19:20 well if the real devices on the target speak scsi, there is an obvious advantage 2007-12-18 19:21 Wait... which layer of FC is it doing? I had assumed it was FC1 or FC2. If it is FC4 than it is, for all intents and purposes, SCSIoE. 2007-12-18 19:21 SCSI/ATA are becoming so similar it is hard to tell them apart. 2007-12-18 19:21 At the protocol level at least. 2007-12-18 19:22 perhaps AoE is getting dated then 2007-12-18 19:22 whats the diff between fc1,2, 4? 2007-12-18 19:22 Possibly. I think it supports scatter/gather and connect/disconnect. 2007-12-18 19:22 Different OSI layers: http://en.wikipedia.org/wiki/Fibre_Channel#Fibre_Channel_layers 2007-12-18 19:25 You know, if you think back to the things that were said about Ethernet back in the day.... it's kind of funny how literally *everything* is moving to it. Ethernet is like the x86 of network platforms. ;-) 2007-12-18 19:33 this whole fcoe things looks a whole heck of a lot more complex than aoe 2007-12-18 19:38 shapor, no, but fiber channel over ethernet sounds even more distasteful than scsi or ata over ethernet 2007-12-18 19:38 somebody will need that solution though 2007-12-18 19:38 somebody with applications that were written for fiber channel and can't be changed for some reason 2007-12-18 19:38 where "some reason" is likely a long, sad tale 2007-12-18 19:39 its good for san admins who want to save money on switches i imagine 2007-12-18 19:40 but buying in to fc just to have an ethernet storage network seems like a lot ot bite off 2007-12-18 19:51 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-18 19:56 so how can i create a file larger than 2TB 2007-12-18 19:58 $ dd if=/dev/zero of=testfile bs=1M seek=2M count=1 2007-12-18 19:58 works 2007-12-18 19:58 $ dd if=/dev/zero of=testfile bs=1M seek=3M count=1 2007-12-18 19:58 dd: truncating at 3298534883328 bytes in output file `testfile': File too large 2007-12-18 20:00 is there still a 2^32*512 bytes file size limit in linux? 2007-12-18 20:08 oh right that is just an ext3 limitation 2007-12-18 20:08 also a limitation of tmpfs 2007-12-18 20:08 so to create a multi-TB sparse file in ext3, I have to create a smaller (1GB) spare file, create an xfs filesystem on it, mount it, and create my sparse file there 2007-12-18 20:09 how silly 2007-12-18 20:09 lots of stacking 2007-12-18 20:19 shapor: that works? 2007-12-18 20:19 yes 2007-12-18 20:19 I guess it would. 2007-12-18 20:20 i just did it 2007-12-18 20:20 Why not just create an XFS filesystem? 2007-12-18 20:20 because this machine has no free space 2007-12-18 20:20 ACTION realizes he just read an e-mail about why. 2007-12-18 20:20 which email? 2007-12-18 20:20 But working with sparse files isn't the same as testing "real" 2TB. 2007-12-18 20:20 the one i sent really isn't related 2007-12-18 20:21 The one talking about the XFS fixes not being in yet. 2007-12-18 20:21 yeah but the reason for me testing that silly way 2007-12-18 20:21 is being on a laptop 2007-12-18 20:21 with no free partition space 2007-12-18 20:21 and an ext3 on the whole thing 2007-12-18 20:21 but it does work 2007-12-18 20:22 $ ./ddsnap status /tmp/server.sock 2007-12-18 20:22 Snapshot store block size = 4096; 262,262 of 262,400 chunks free 2007-12-18 20:22 Origin size: 5497559187456 bytes 2007-12-18 20:22 we need the pretty-print for origin size :) 2007-12-18 20:22 hehe 2007-12-18 20:24 hrm actually it is in 2007-12-18 20:24 ah 2007-12-18 20:24 is my large number breaking the pretty-print? 2007-12-18 20:24 perhaps its too long for the char buf 2007-12-18 20:25 lol, that'd be awesome if that was the case. 2007-12-18 20:25 it is 2007-12-18 20:25 16 chars 2007-12-18 20:25 that rocks 2007-12-18 20:25 My guess is that we're sucking it in to too small a char. 2007-12-18 20:25 at least it fails gracefully 2007-12-18 20:25 s/char/int/ 2007-12-18 20:25 yeah 2007-12-18 20:25 no 2007-12-18 20:26 it prints out fine, just not pretty 2007-12-18 20:26 nice 2007-12-18 20:26 commas() fails gracefully though 2007-12-18 20:26 thanks snprintf 2007-12-18 20:26 A key aspect of using snprintf is computing what the maximum should be, or resizing the buffer it is exceeded. 2007-12-18 20:26 1819 char number1[16], number2[16]; 2007-12-18 20:26 such meaningful identifiers 2007-12-18 20:27 ACTION rolls eyes 2007-12-18 20:27 awesomeness 2007-12-18 20:27 There are a lot of parts to ddsnap that just make me cringe. 2007-12-18 20:28 $ ./ddsnap status /tmp/server.sock 2007-12-18 20:28 Snapshot store block size = 4096; 262,262 of 262,400 chunks free 2007-12-18 20:28 Origin size: 5,497,559,187,456 bytes 2007-12-18 20:28 ah thats btter 2007-12-18 20:28 ACTION looks at the code and cries 2007-12-18 20:29 You can actually compute the maximum size needed for those buffers... why pick 16? 2007-12-18 20:29 s/cries/fixes it/ 2007-12-18 20:29 would be better :) 2007-12-18 20:32 did you set the size to 21 bytes? 2007-12-18 20:32 yes 2007-12-18 20:32 kk 2007-12-18 20:32 Just wanted to avoid visiting that particular avenue again. 2007-12-18 20:33 We should probably do "#define MAX_LONG_LONG_STRING_LENGTH 21" somewhere 2007-12-18 20:33 replace 2 characters with 27? 2007-12-18 20:34 Well, I can imagine something terser, but most importantly, replacing a "magic number" with a "name with describes how it was selected". 2007-12-18 20:35 "#define U64_STR_LENGTH 21" is reasonably terse and still conveys the how and the why of it. 2007-12-18 20:35 Actually, I guess it is just INT64. U64 would be 20 bytes. 2007-12-18 20:36 its not long enough 2007-12-18 20:36 oh yeah.. the commas. 2007-12-18 20:36 drat. 2007-12-18 20:36 also commas() takes a long long, not unsigned long long 2007-12-18 20:36 perhaps a negative sign as well 2007-12-18 20:36 yeah, hence why I said "int64". 2007-12-18 20:36 but the format string uses %Lu, hm 2007-12-18 20:37 yeah but you need a null at the end too 2007-12-18 20:37 21 is enough for any int64... without commas. 2007-12-18 20:38 MAX_LONG_LONG is 19 digits long. I added one more byte for the negative and another for the null. That's how I got 21. 2007-12-18 20:38 ACTION can see why it makes sense to just have the compiler compute it, so as to avoid this discussion again. ;-) 2007-12-18 20:38 So, you need to add 19/3 = 6 more bytes for the commas. 2007-12-18 20:39 27 is the number for unsigned 2007-12-18 20:39 i guess signed is the same 2007-12-18 20:39 since you gain a -, but lose a digit 2007-12-18 20:40 lol, commas actually computes how much you need. We should just break that out in to a separate function and then have a static const that stores the result for calling it with MAX_LONG_LONG. 2007-12-18 20:40 yup 2007-12-18 20:40 I forgot you gained another digit. 2007-12-18 20:41 it computes it for your particular value 2007-12-18 20:41 its silly to break it out in to another function 2007-12-18 20:41 just add a comment, max required string len is 27 :) 2007-12-18 20:44 static size_t bytes_needed_for_commas(long long n) 2007-12-18 20:44 { 2007-12-18 20:44 int newlen = snprintf(0, 0, "%Lu", n); 2007-12-18 20:44 if (newlen == 0) { 2007-12-18 20:44 return 0; 2007-12-18 20:44 } 2007-12-18 20:44 return newlen + ((newlen-1)/3); 2007-12-18 20:44 } 2007-12-18 20:44 const size_t U64_BUF_LENGTH = bytes_needed_for_commans(MAX_LONG_LONG); 2007-12-18 20:44 why compute something that never changes? 2007-12-18 20:44 It only computes it once. 2007-12-18 20:45 every time you run the program 2007-12-18 20:45 on every machine its installed on 2007-12-18 20:45 And you compute it because computers are reliable at arithmetic. People are not. 2007-12-18 20:45 compute it once and store it 2007-12-18 20:45 it takes more storage to store that computation than the number 2007-12-18 20:46 You do the computation in commas anyway. 2007-12-18 20:46 Sadly in C++ you could get it all evaluated at compile time... actually, maybe C99 has the value in one of its headers.... 2007-12-18 20:49 18 exabytes, hrm 2007-12-18 20:50 Actually, I guess a sufficiently clever C compiler would still compute this at compile time. 2007-12-18 20:52 including a call to snprintf() ? 2007-12-18 20:52 *highly* unlikely 2007-12-18 20:53 A clever compiler would recognize that the snprintf() is always invoked with the same constants and therefore will always return the same results. 2007-12-18 20:53 have you looked at stdio code? heh 2007-12-18 20:53 It'd help if I hinted a bit more by making newlen const, but most compilers will figure that out for themselves anyway. 2007-12-18 20:54 Yeah, that's why I qualified it with "sufficiently clever" ;-) 2007-12-18 20:54 smarter than you or i 2007-12-18 20:54 the other way to compute it is to count bits. 2007-12-18 20:54 at that point just tellthe compiler you want a snapshot server 2007-12-18 20:54 and then invoke log() ;-) 2007-12-18 20:54 a sufficiently clever one would plop one out 2007-12-18 20:54 lol 2007-12-18 20:55 Yes, I was deliberately referencing the fallacy of the sufficiently clever compiler. ;-) irc.oftc.net #zumastor log beginning Wed Dec 19 00:00:01 PST 2007 2007-12-19 00:02 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-19 01:16 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-19 03:00 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 07:34 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-19 08:11 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2007-12-19 08:28 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 09:55 good morning 2007-12-19 09:56 morning... 2007-12-19 09:59 http://blogs.cnet.com/8301-13505_1-9835693-16.html?part=rss&subj=news&tag=2547-1_3-0-5 2007-12-19 09:59 thanks for the link tim_vimm :-) 2007-12-19 10:00 my pleasure :) 2007-12-19 10:03 time to get on the scramjet shoes 2007-12-19 10:03 hmm, maybe will need some underwater breathing apparatus as well 2007-12-19 10:04 at least a wet suit 2007-12-19 10:04 surf on over 2007-12-19 10:04 give the scramjet shoes a day off 2007-12-19 10:07 scramjet surfboard 2007-12-19 10:07 the electric skateboard would short out today 2007-12-19 12:56 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-19 14:17 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 14:53 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-19 15:11 -!- pgquiles__(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-19 16:14 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 16:14 -!- kconstan_(~kconstan@wdas-1.disneyanimation.com) has joined #zumastor 2007-12-19 16:34 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 16:42 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 16:49 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 19:06 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 19:33 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 21:36 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 21:40 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 22:50 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 22:55 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-19 23:36 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Thu Dec 20 00:00:01 PST 2007 2007-12-20 04:37 -!- dank(~chatzilla@cpe-76-90-56-73.socal.res.rr.com) has joined #zumastor 2007-12-20 08:13 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-20 10:35 good morning 2007-12-20 10:38 Good morning... 2007-12-20 10:38 Ran into a Digital Domain dude last night 2007-12-20 10:39 said he saw the lkml post 2007-12-20 10:39 future of storage? 2007-12-20 10:39 yup 2007-12-20 10:39 said "you guys got a nice write up" 2007-12-20 10:40 so we know of at least one reader :P 2007-12-20 10:40 from that one data point, the exposure seems pretty wide 2007-12-20 10:40 that's not the only data point though 2007-12-20 10:40 it would seem 2007-12-20 10:40 there will be a sequel 2007-12-20 10:40 the other(s)? 2007-12-20 10:41 ah 2007-12-20 10:41 couple of googlers mentioned the post, I had no idea they followed lkml 2007-12-20 10:41 not kernel people 2007-12-20 10:41 I'm now a regular reader 2007-12-20 10:41 thanks 2 you 2007-12-20 10:41 :) 2007-12-20 10:42 the sequel fill push the throughput tests up until latency goes to 1000 ms or so 2007-12-20 10:42 and I hope to know more about how fstress results relate to spec sfs 2007-12-20 10:43 lkml can rot your brain ;) 2007-12-20 10:43 I'm not even a regular reader 2007-12-20 10:43 rotting away as we speak 2007-12-20 10:43 just got a private email from scotland about the other post 2007-12-20 10:44 good/bad/ugly? 2007-12-20 10:44 asking questions, so good 2007-12-20 10:44 people seem shy about commenting on this subject 2007-12-20 10:45 most likely to avoid being flamed 2007-12-20 10:45 for one thing, they may not have ever seen the problem because of chicken and egg: if linux sucks for storage then nobody will run storage apps on linux that break it 2007-12-20 11:26 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2007-12-20 11:38 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-20 11:39 flips: ping 2007-12-20 11:39 shapor, pong 2007-12-20 11:40 so, according to drake, sync isnt enough on xfs 2007-12-20 11:41 by what logic? 2007-12-20 11:42 as in, create a file, sync, take a snapshot, file is not in snapshot 2007-12-20 11:42 I would like to see that with my own eyes 2007-12-20 11:42 would be a posix violation 2007-12-20 11:43 yeah i'm going to try it 2007-12-20 11:43 manually 2007-12-20 11:44 #if the VFS lock patch is present, the calls to xfs_freeze are redundant, but should cause no problems 2007-12-20 11:44 hm from http://linux-xfs.sgi.com/projects/xfs/mail_archive/200208/msg00576.html 2007-12-20 11:44 we are still waiting for a fix to the dm-ddsnap shfit count, right? 2007-12-20 11:45 they also call sync 2007-12-20 11:46 we don't even require lock_fs 2007-12-20 11:47 I forget danks argument why we should make sure snapshot without lock_fs works properly, it made sense at the time 2007-12-20 11:48 the idea of being able to take a snapshot just whenever, without flushing things, sending meesages to applications etc has always been appealing to me 2007-12-20 11:48 it's like the idea of being able to shut the power off on your workstation and have it all start up properly again 2007-12-20 11:48 it is supposed to work 2007-12-20 11:48 i do that all the time 2007-12-20 11:49 works for me 2007-12-20 11:49 usually due to laptop battery dying when i fall asleep :) 2007-12-20 11:49 dank's presented his argument around the time we invented the odd even snapshot concept 2007-12-20 11:49 seemed worth remembering... 2007-12-20 11:49 i dont remember 2007-12-20 11:50 the question is why does xfs_freeze exist then, and what does it actually do 2007-12-20 11:51 vestigial hangover from before there was a lock_fs? 2007-12-20 11:51 because freeze existed on sgi when xfs was ported? 2007-12-20 11:54 is that different than blockdev --flushbufs 2007-12-20 11:59 23 #define BLKFLSBUF _IO(0x12,97) 2007-12-20 11:59 from blockdev.c 2007-12-20 12:00 is what different? 2007-12-20 12:00 lock_fs 2007-12-20 12:00 i thought lock_fs was only callable from dm 2007-12-20 12:01 BLKFLSBUF may not act on the filesystem, only the block device 2007-12-20 12:01 need to dive into the code to know 2007-12-20 12:01 I wonder when lxr.linux.no is going to fix their style sheet 2007-12-20 12:02 yeah the blockdev --flushbufs we use to ensure any pending ios are completed before using our snapshot (to close the race) is just an ioctl interface 2007-12-20 12:02 not sure what it translates to 2007-12-20 12:02 i was thinking of running lxr on zumastor.org 2007-12-20 12:02 it seems to be the new gen lxr 2007-12-20 12:02 or is it just an improvement of 0.3, not building on the 0.9 stuff 2007-12-20 12:02 do a service for the community and be able to browse our patched repo 2007-12-20 12:03 how about when I put up my dd tree? 2007-12-20 12:03 then lxr that 2007-12-20 12:03 otherwise, lxr the ddsnap userspace code? 2007-12-20 12:03 that would be useful 2007-12-20 12:03 and much less work for the server 2007-12-20 12:03 i was thinking all kernel versions 2007-12-20 12:04 lxr is a server/bandwith resource suck 2007-12-20 12:05 # lock_kernel(); 2007-12-20 12:05 # fsync_bdev(bdev); 2007-12-20 12:05 # invalidate_bdev(bdev); 2007-12-20 12:05 # unlock_kernel(); 2007-12-20 12:05 # return 0; 2007-12-20 12:05 that is what BLKFLSBUF does 2007-12-20 12:05 just flush the device 2007-12-20 12:06 that is in fact all we need 2007-12-20 12:06 so sync is overkill 2007-12-20 12:06 and freeze_bdev is way overkill 2007-12-20 12:07 well good because BLKFLSBUF is exactly what we do ;) 2007-12-20 12:07 and lock_fs is a self-inflicted nuke 2007-12-20 12:07 oh good, we don't suck 2007-12-20 12:07 xfs must be cheating 2007-12-20 12:07 that would explain why sync doesn't work 2007-12-20 12:07 xfs is not allowed to cheat 2007-12-20 12:08 I think the observer may have erred 2007-12-20 12:08 well i shall head in to the office and check then 2007-12-20 12:08 ACTION doesn't feel like vpn'ing in 2007-12-20 12:08 oh you finally crossed that threshold 2007-12-20 12:09 well now that zumastor is a production service at home at cant just mess with that machine 2007-12-20 12:09 i have to use machines at work :) 2007-12-20 12:09 it is? 2007-12-20 12:09 beat me too it 2007-12-20 12:09 to it even 2007-12-20 12:09 yeah music is playing from it right now 2007-12-20 12:10 so I'm playing catchup 2007-12-20 12:10 all because I couldn't get new hard disks from newegg fast enough ;) 2007-12-20 12:10 mines been setup a while 2007-12-20 12:10 crappy old 400GB PATA 2007-12-20 12:11 most of my machines here are still pata 2007-12-20 12:11 only two satas 2007-12-20 12:11 probably similar to the rest of the world 2007-12-20 12:11 with a really old 40GB ibm as snapstore 2007-12-20 12:11 40 gb :) 2007-12-20 12:12 well good enough for uploading photos 2007-12-20 12:12 yeah, its a second spindle 2007-12-20 12:12 snapshots get squashed when you upload too many movies :) 2007-12-20 12:13 the nice thing about our snapshot rotation is that 2007-12-20 12:13 I could stick in a 2.5 disk for snapshot store, nice and quiet 2007-12-20 12:13 if the server is off for a week, i still have the same number of snapshots 2007-12-20 12:13 better than setting expire dates for them 2007-12-20 12:13 very much so 2007-12-20 12:14 now I must re-merge the ddsnapd revert patch, somebody has been fiddling with error messages in my absense 2007-12-20 12:14 now fails 2007-12-20 12:14 :P 2007-12-20 12:14 i didnt commit that 2007-12-20 12:14 well it happens 2007-12-20 12:14 wait, someone did? 2007-12-20 12:14 out of tree patches are bad 2007-12-20 12:15 I should simply have merged it at the time 2007-12-20 12:15 I won't make the same mistake again, when I port it this time I will just merge it and we can test it along with other 0.5 code 2007-12-20 12:15 just commit I mean 2007-12-20 12:16 if it breaks anything, the breakage will not be subtle 2007-12-20 12:16 maybe post a patch first :) 2007-12-20 12:16 already did 2007-12-20 12:16 you did? 2007-12-20 12:16 yes 2007-12-20 12:17 ok, I will freshen the patch an post to googlegroups 2007-12-20 12:17 hm maybe i should commit my error message patch before you freshen it 2007-12-20 12:18 otherwise i'll have to freshen that 2007-12-20 12:18 why has braindamaged .rej format never been improved in all these years? 2007-12-20 12:18 is there some subtle beauty about it I have missed? 2007-12-20 12:19 because only programming rejects get rej's :P 2007-12-20 12:19 oh that is what I missed 2007-12-20 12:21 actually the error handler patch shouldnt conflict with yours, so dont worry about it 2007-12-20 12:22 i found your original patch.. forgot about it.. i even commented on it 2007-12-20 12:23 yet another example of how waiting before committing causes more work than committing early, then picking up the broken pieces 2007-12-20 12:23 "commit early, commit often" 2007-12-20 12:24 oh trunk anyway 2007-12-20 12:24 on I mean 2007-12-20 12:25 now what I need to do is find the original file I diffed against (rev 931) and diff it against the contents of the --- part of .rej, with the leading character of each line removed 2007-12-20 12:25 ACTION kicks flips to too at the error handler patch 2007-12-20 12:25 sounds like a generic, useful opperation 2007-12-20 12:26 s/too/look/ 2007-12-20 12:26 ok 2007-12-20 12:26 I didn't like the cut n paste itis I saw in somebody else's error cleanups 2007-12-20 12:26 I assume you also noticed 2007-12-20 12:26 yeah dank's patch, i got rid of that 2007-12-20 12:26 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-20 12:27 I don't see your patch on gogglegroups yet 2007-12-20 12:27 thats why i did the cleanups 2007-12-20 12:28 http://groups.google.com/group/zumastor/browse_thread/thread/75e42fe0f1408416# 2007-12-20 12:28 ok how do I get svn to give me a copy of a file as at a given rev? 2007-12-20 12:28 svn update -r 123 filename 2007-12-20 12:28 i think 2007-12-20 12:29 doesn't that change the file? 2007-12-20 12:29 or perhaps revert not update 2007-12-20 12:29 well 2007-12-20 12:29 lame 2007-12-20 12:30 what I want is svn cat + -r 2007-12-20 12:30 would seem to be a fundamental operation 2007-12-20 12:30 svn cat -r 123 filename 2007-12-20 12:31 cat doesn't support gnu opts? 2007-12-20 12:31 svn cat certainly could 2007-12-20 12:31 should even 2007-12-20 12:31 oh 2007-12-20 12:31 there it is 2007-12-20 12:32 maybe svn cat --cat-cmd cat -x :) 2007-12-20 12:32 like diff :) 2007-12-20 12:33 it is not clear to me why it is not just --diff-cmd "command name with opts" 2007-12-20 12:33 or even "command name with opts or just opts" 2007-12-20 12:34 the answer to all your git questions is here http://video.google.com/videoplay?docid=-2199332044603874737 2007-12-20 12:34 s/git/svn/ 2007-12-20 12:35 :P 2007-12-20 12:42 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-20 13:00 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-20 13:04 "Samba Team Receives Microsoft Protocol Documentation" nttp://www.linuxtoday.com/infrastructure/2007122002526NWMSLL :-) 2007-12-20 13:17 ACTION equips his scramjet shoes 2007-12-20 17:59 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-20 18:01 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-20 22:05 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2007-12-20 22:16 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-20 23:59 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Fri Dec 21 00:00:01 PST 2007 2007-12-21 01:04 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-21 02:31 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2007-12-21 08:20 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-21 10:28 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-21 10:54 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-21 11:11 http://www.redhat.com/archives/ext3-users/2005-March/msg00025.html - somebody proposing using freeze_bdev / snapshot / thaw_bdev 2007-12-21 11:11 on ext3 2007-12-21 11:48 it is the "right" think to do 2007-12-21 11:48 however snapshotting without freeze ought to work 2007-12-21 11:49 and has the definite advantage of no loss of service during the freeze 2007-12-21 11:54 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-21 12:03 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-21 14:02 -!- pgquiles(~pgquiles@100.Red-83-33-71.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-21 14:24 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-21 14:26 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-21 16:10 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-21 18:20 -!- Tim_vimm(~Tim@rrcs-64-183-50-58.west.biz.rr.com) has joined #zumastor 2007-12-21 19:13 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-21 19:30 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-21 21:14 -!- cbsmith(~xman@adsl-76-243-138-218.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-21 21:27 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sat Dec 22 00:00:01 PST 2007 2007-12-22 03:47 -!- pgquiles(~pgquiles@100.Red-83-33-71.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-22 08:10 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-22 09:53 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-22 11:51 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-22 13:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-22 14:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-22 16:47 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-22 17:52 -!- pgquiles(~pgquiles@174.Red-81-36-252.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-22 18:51 hi all 2007-12-22 18:51 quiet day here 2007-12-22 20:00 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-22 21:03 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sun Dec 23 00:00:01 PST 2007 2007-12-23 02:20 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2007-12-23 02:23 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2007-12-23 08:14 -!- erwan_taf(~erwan@90.12.47.252) has joined #zumastor 2007-12-23 08:52 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-23 09:23 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-23 11:13 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-23 11:50 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-23 14:27 -!- pgquiles(~pgquiles@120.Red-81-35-183.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-23 15:11 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-23 18:34 -!- willn(~wan@paperboy.ccs.neu.edu) has joined #zumastor 2007-12-23 18:58 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-23 19:42 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-23 23:02 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Dec 24 00:00:01 PST 2007 2007-12-24 05:57 -!- pgquiles(~pgquiles@120.Red-81-35-183.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-24 08:46 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-24 09:21 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-24 10:17 -!- pgquiles(~pgquiles@120.Red-81-35-183.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-24 10:21 -!- pgquiles_(~pgquiles@186.Red-81-32-4.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-24 12:08 -!- pgquiles(~pgquiles@186.Red-81-32-4.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-24 13:12 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-24 14:58 -!- pgquiles(~pgquiles@186.Red-81-32-4.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-24 15:01 -!- pgquiles(~pgquiles@186.Red-81-32-4.dynamicIP.rima-tde.net) has joined #zumastor irc.oftc.net #zumastor log beginning Tue Dec 25 00:00:01 PST 2007 2007-12-25 03:22 -!- pgquiles(~pgquiles@165.Red-81-38-97.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-25 04:00 -!- pgquiles_(~pgquiles@156.Red-83-41-112.dynamicIP.rima-tde.net) has joined #zumastor 2007-12-25 14:24 ACTION looks around for any sign of life 2007-12-25 16:13 -!- pgquiles(~pgquiles@156.Red-83-41-112.dynamicIP.rima-tde.net) has joined #zumastor irc.oftc.net #zumastor log beginning Wed Dec 26 00:00:01 PST 2007 2007-12-26 08:22 -!- jinevill(~justin@user-0c99kat.cable.mindspring.com) has joined #zumastor 2007-12-26 08:22 -!- jinevill(~justin@user-0c99kat.cable.mindspring.com) has left #zumastor 2007-12-26 11:28 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-26 11:39 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-26 12:39 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-26 17:19 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-26 17:20 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2007-12-26 19:02 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-26 20:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2007-12-26 20:45 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor irc.oftc.net #zumastor log beginning Thu Dec 27 00:00:01 PST 2007 2007-12-27 01:36 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2007-12-27 09:10 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-27 10:26 ACTION looks around for signs of life 2007-12-27 10:26 so quiet... 2007-12-27 11:14 yup, it's still the holidays for most people 2007-12-27 12:13 well at least we're around, the two with 0.5 bugs assigned 2007-12-27 12:13 so we dont get a break 2007-12-27 12:14 i'm going to look in to the test brokenness from drakes changes 2007-12-27 12:14 2tb does indeed work 2007-12-27 12:14 so my issue should be closable 2007-12-27 12:31 -!- SEJeff(~jeff@office4.tmcs.net) has joined #zumastor 2007-12-27 12:41 I think I can get mine closed today as well. Mostly I just need to make sure they are suitable for peer review. 2007-12-27 12:45 send them to the list when ready and I can review 2007-12-27 12:49 kk 2007-12-27 19:35 hey, when did INPUT_ERROR show up? 2007-12-27 19:52 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-27 22:44 cbsmith: not too long ago 2007-12-27 23:09 shapor: Well, as you can see I submitted a patch to "fix" it. irc.oftc.net #zumastor log beginning Fri Dec 28 00:00:01 PST 2007 2007-12-28 11:22 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2007-12-28 12:16 cbsmith: around? 2007-12-28 22:58 -!- flipz(~chatzilla@207.102.153.42) has joined #zumastor 2007-12-28 23:01 ACTION gives chatzilla a try irc.oftc.net #zumastor log beginning Sat Dec 29 00:00:01 PST 2007 2007-12-29 02:16 -!- lili20(~lili20@ANantes-257-1-162-177.w86-195.abo.wanadoo.fr) has joined #zumastor 2007-12-29 15:27 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor irc.oftc.net #zumastor log beginning Sun Dec 30 00:00:01 PST 2007 irc.oftc.net #zumastor log beginning Mon Dec 31 00:00:01 PST 2007 2007-12-31 19:44 -!- flips(~phillips@phunq.net) has joined #zumastor