irc.oftc.net #zumastor log beginning Tue Jan 1 00:00:01 PST 2008 2008-01-01 01:58 -!- pgquiles(~pgquiles@192.Red-83-34-135.dynamicIP.rima-tde.net) has joined #zumastor 2008-01-01 01:59 -!- pgquiles(~pgquiles@192.Red-83-34-135.dynamicIP.rima-tde.net) has joined #zumastor 2008-01-01 03:05 -!- Guest1843(~knoppix@202.141.47.33) has joined #zumastor 2008-01-01 03:05 hi 2008-01-01 06:14 -!- pgquiles(~pgquiles@82.Red-83-49-101.dynamicIP.rima-tde.net) has joined #zumastor 2008-01-01 12:24 happy new year! 2008-01-01 12:35 happy new year indeed! 2008-01-01 12:35 been a quiet week on here 2008-01-01 12:38 partly because I was incommunicado 2008-01-01 12:38 stuck behind a 56k dialup windows machine 2008-01-01 12:40 eww 2008-01-01 12:41 people still use modems? 2008-01-01 12:41 when they can't get dsl or cable what's the alternative? 2008-01-01 12:41 satellite? ++lag 2008-01-01 12:42 i've been spoiled with cable since 1996 2008-01-01 12:43 -!- pgquiles(~pgquiles@81.202.65.108) has joined #zumastor 2008-01-01 12:43 3g or edge perhaps, although thats lag+ as well 2008-01-01 12:45 isn't that a little expensive? 2008-01-01 12:47 i suppose it is compared to dial up 2008-01-01 12:51 well I must get an edge capable phone 2008-01-01 12:52 so I don't get stuck in the boondocks again 2008-01-01 12:52 charges will be scary though 2008-01-01 12:52 especially if roaming 2008-01-01 12:52 let's see, about $120/hour 2008-01-01 12:53 I hope I'm wrong 2008-01-01 22:58 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Wed Jan 2 00:00:01 PST 2008 2008-01-02 08:42 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-02 09:16 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-02 09:21 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-02 09:32 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-02 09:48 -!- kconstan_(~kconstan@wdas-1.disneyanimation.com) has left #zumastor 2008-01-02 11:14 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2008-01-02 12:36 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-02 14:02 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-02 15:45 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-02 18:55 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-02 20:44 -!- clara30(~clara30@ANantes-257-1-162-177.w86-195.abo.wanadoo.fr) has joined #zumastor irc.oftc.net #zumastor log beginning Thu Jan 3 00:00:01 PST 2008 2008-01-03 00:11 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-03 01:23 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-03 07:08 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-03 09:33 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-03 10:13 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-03 10:20 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-03 13:55 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-03 15:44 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-03 18:03 -!- cbsmith(~xman@adsl-75-56-192-117.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-03 18:37 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-03 20:07 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-03 23:44 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Fri Jan 4 00:00:01 PST 2008 2008-01-04 10:00 -!- SEJeff(~jeff@office4.tmcs.net) has joined #zumastor 2008-01-04 11:26 -!- pgquiles(~pgquiles@171.Red-217-125-197.dynamicIP.rima-tde.net) has joined #zumastor 2008-01-04 20:57 You guys presenting at SCALE again this year? 2008-01-04 21:11 nope 2008-01-04 21:11 i submitted a paper but it didn't get selected 2008-01-04 21:11 well an abstract at least 2008-01-04 21:14 it was very last minute 2008-01-04 21:17 attending then? 2008-01-04 21:38 yeah i'm planning on going 2008-01-04 21:38 its just down the street really 2008-01-04 21:39 also planning on attending lsf and fast in feb 2008-01-04 22:18 -!- nataliep_(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-04 22:20 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-04 23:28 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sat Jan 5 00:00:01 PST 2008 2008-01-05 03:50 -!- pgquiles(~pgquiles@171.Red-217-125-197.dynamicIP.rima-tde.net) has joined #zumastor 2008-01-05 13:35 -!- nataliep_(~nataliep@66-102-14-1.google.com) has joined #zumastor 2008-01-05 13:35 -!- jiayingz(~jiayingz@66-102-14-1.google.com) has joined #zumastor 2008-01-05 15:01 -!- dank(~chatzilla@adsl-69-235-31-113.dsl.irvnca.pacbell.net) has joined #zumastor 2008-01-05 20:42 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2008-01-05 20:59 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2008-01-05 20:59 we were offline for a couple hours 2008-01-05 20:59 hardware issues 2008-01-05 21:00 zumastor.org back up and running irc.oftc.net #zumastor log beginning Sun Jan 6 00:00:01 PST 2008 2008-01-06 13:02 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-06 21:01 hey shapor 2008-01-06 23:49 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Jan 7 00:00:01 PST 2008 2008-01-07 02:11 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-07 02:11 hi flips 2008-01-07 09:53 good morning shapor 2008-01-07 09:54 it's scramjet time 2008-01-07 10:04 indeed, i just scramed in 2008-01-07 10:44 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2008-01-07 10:44 ACTION is delayed at two airports 2008-01-07 10:45 will get to the westcoast sometime 2008-01-07 10:50 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-07 11:37 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-07 12:10 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-07 16:01 -!- cbsmith(~user@adsl-71-133-80-65.dsl.irvnca.pacbell.net) has joined #zumastor 2008-01-07 16:03 ACTION reappears 2008-01-07 16:04 ACTION pings shapor 2008-01-07 16:19 hey nataliep_ 2008-01-07 16:22 wow, the new lxr.linux.no is really much better than the old, in spite of ugly interface 2008-01-07 16:22 search is excellent and instant 2008-01-07 16:25 flips: That's just 'cause noone wants to use it. :-) 2008-01-07 16:25 ? 2008-01-07 16:25 ugly interface == fewer users == great responsiveness :-) 2008-01-07 16:26 the old one sucked even in the middle of the night 2008-01-07 16:26 performance wise 2008-01-07 16:26 they eventually just disabled the search 2008-01-07 16:27 Think of it as a new take on "worse is better" ;-) 2008-01-07 16:27 The "Software As A Service" version 2008-01-07 16:28 unfortunately, the new version produces wrong results 2008-01-07 16:29 try a code search on throttle_vm_writeout 2008-01-07 16:29 correct result is: 2008-01-07 16:29 $ grep throttle_vm_writeout * -r 2008-01-07 16:29 include/linux/writeback.h:void throttle_vm_writeout(gfp_t gfp_mask); 2008-01-07 16:29 mm/page-writeback.c:void throttle_vm_writeout(gfp_t gfp_mask) 2008-01-07 16:29 mm/vmscan.c: throttle_vm_writeout(sc->gfp_mask); 2008-01-07 16:30 lxr.linux.no result is (http://lxr.linux.no/linux+v2.6.23.12/+ident=3548186) 2008-01-07 16:30 Function at mm/page-writeback.c, line 329 References: mm/page-writeback.c, line 329 2008-01-07 16:31 flips: Huh... I got a different result set when I searched 2008-01-07 16:31 ok, the result is limited to the directory you have selected 2008-01-07 16:31 that is nice 2008-01-07 16:31 :) 2008-01-07 16:31 Ah, that'd be what it was. 2008-01-07 16:31 I also got a hit for include/linux/writeback.h 2008-01-07 16:33 cbsmith hi 2008-01-07 16:33 gee, congestion_wait and friends has become even more of a sprawling mess than I thought 2008-01-07 16:34 shapor: So, after talking with DanK I realized you guys were passing this patch around because I'd failed to submit mine. 2008-01-07 16:34 ah 2008-01-07 16:34 i was wondering about your question 2008-01-07 16:34 shapor: Anyway, I was going to ask you why we didn't like doing it inline. 2008-01-07 16:34 we had talked about it xmas week 2008-01-07 16:34 inline? 2008-01-07 16:35 shapor: Yup, took me a while to figure out how to get the diff right (svn add & remove is the only way to tell svn diff to show files getting added and removed). 2008-01-07 16:35 what did you do that added/removed files? 2008-01-07 16:35 Yeah, I had a script that copied it over in to /sbin, and in the process prepended "SVNVERSION=`svnversion.....`" to the beginning of the zumastor script. 2008-01-07 16:35 ah 2008-01-07 16:35 yeah we toyed with a simlar idea 2008-01-07 16:35 I moved the zumastor script out of /bin 2008-01-07 16:36 but didn't like the idea of modifying the script 2008-01-07 16:36 because it might sneak in to a commit 2008-01-07 16:36 Well, I wasn't modifying it in place, which I think makes it just fine. 2008-01-07 16:36 is the prevailing idea still to have /etc/zumastor/version ? 2008-01-07 16:36 flips yes 2008-01-07 16:36 Yeah, mine couldn't accidentally make it in to a commit. 2008-01-07 16:36 The version controlled version of the file doesn't get modified. 2008-01-07 16:37 Does that perhaps overcome the "store the version inline" objections? 2008-01-07 16:38 If it might, I'm happy to post it for further review. 2008-01-07 16:38 shapor, and in the source tree would the logical place to store the zumastor version be trunk/zumastor/etc/version? 2008-01-07 16:39 flips: I should think it wouldn't be logical to store the zumastor version in the source tree, but rather to compute it on the fly. 2008-01-07 16:39 flips: Just like we do for ddsnap 2008-01-07 16:39 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-07 16:39 howdy 2008-01-07 16:40 cbsmith, perhaps you mean the svn commit number? 2008-01-07 16:40 shapor, what would cat /etc/zumastor/version give? 2008-01-07 16:41 the exact same output as zumastor --version 2008-01-07 16:41 which would be? 2008-01-07 16:41 i dunno some long thing 2008-01-07 16:41 same format as ddsnap --version 2008-01-07 16:41 we can change it if we want 2008-01-07 16:41 how about an example 2008-01-07 16:41 ACTION tries ddsnap --version 2008-01-07 16:42 whoa, that is ugly 2008-01-07 16:42 why is there a leading space on the second line, and why two lines? 2008-01-07 16:43 why are there quotes around the revision number? 2008-01-07 16:43 i got rid of the newline 2008-01-07 16:43 i think that was a typo 2008-01-07 16:43 and i got rid of the quotes too 2008-01-07 16:43 zumastor revision 1247M built on Mon Jan 7 16:42:35 PST 2008 by shapor@shapor 2008-01-07 16:43 :) 2008-01-07 16:43 where is the "0.4" ? 2008-01-07 16:43 not in there 2008-01-07 16:43 right 2008-01-07 16:43 that is a zumastor thing 2008-01-07 16:43 because that is up a level 2008-01-07 16:44 it needs to get copied in to zumastor and ddsnap directories 2008-01-07 16:44 just liek the SVNREV does 2008-01-07 16:44 when tarballs are rolled 2008-01-07 16:44 ok, I don't really care about seeing 0.4 anywhere just now 2008-01-07 16:44 by the build script 2008-01-07 16:44 yeah i didnt see it as a big issue 2008-01-07 16:45 ok, I will dive back into the congestion_wait mess 2008-01-07 16:45 thus reassured 2008-01-07 16:45 ACTION hears a splash 2008-01-07 16:45 belly flop eh? 2008-01-07 16:45 toxic waste flop 2008-01-07 16:57 Ugh... everyone hates my patch. Ah well. irc.oftc.net #zumastor log beginning Tue Jan 8 00:00:01 PST 2008 2008-01-08 00:10 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-08 01:04 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-08 08:05 -!- charlesnw(~charles@cpe-75-84-73-136.socal.res.rr.com) has joined #zumastor 2008-01-08 11:56 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-08 12:29 -!- charlesnw(~charles@cpe-75-84-73-136.socal.res.rr.com) has left #zumastor 2008-01-08 14:01 -!- SEJeff(~jeff@office4.tmcs.net) has joined #zumastor irc.oftc.net #zumastor log beginning Wed Jan 9 00:00:01 PST 2008 2008-01-09 00:09 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-09 08:40 good morning 2008-01-09 10:01 -!- erwan_taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2008-01-09 10:12 morning flips 2008-01-09 10:12 scramjet time, just about 2008-01-09 11:10 -!- phillips_(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-09 11:30 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-09 14:10 -!- erwan__taf(~erwan@konilope.linuxeries.org) has joined #zumastor 2008-01-09 15:16 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-09 19:28 ACTION just did a run of uml/demo.sh on a vanilla gutsy laptop irc.oftc.net #zumastor log beginning Thu Jan 10 00:00:01 PST 2008 2008-01-10 00:33 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-10 01:25 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-10 02:09 -!- ejt(~ejt@85-211-72-236.dyn.gotadsl.co.uk) has joined #zumastor 2008-01-10 02:09 ACTION waves 2008-01-10 02:30 -!- ejt(~ejt@85-211-72-236.dyn.gotadsl.co.uk) has joined #zumastor 2008-01-10 06:50 -!- ejt(~ejt@85-211-72-236.dyn.gotadsl.co.uk) has joined #zumastor 2008-01-10 07:07 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-10 07:12 flips ? 2008-01-10 09:56 willn: howd that work out? 2008-01-10 10:21 shapor: Was a good way to kill the battery while I ate dinner :D 2008-01-10 10:21 But it reported PASS at completion, I was looking for some more information out of it 2008-01-10 10:45 What kind of data are we looking for out of a performance test lab? 2008-01-10 10:51 I think we understand how we perform, we have a queue of optimizations we want to do to improve performance 2008-01-10 11:17 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-10 12:00 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-10 12:10 hi ejt 2008-01-10 12:10 joe I presume? 2008-01-10 12:43 flips: is there a release date for 0.5? 2008-01-10 12:43 pgquiles: today :) 2008-01-10 12:44 http://zumastor.org/downloads/releases/0.5-r1264/ 2008-01-10 12:44 shapor: cool :-) 2008-01-10 12:45 funny you should ask ;) 2008-01-10 12:46 0.6 opening in a couple days is defined as the release where we clean up lots of things 2008-01-10 12:47 say, are googlers outnumbered by nongooglers here today? 2008-01-10 12:47 it's close anyway 2008-01-10 12:47 looking forward to being outnumbered 2008-01-10 12:48 what's the status of remote replication + snapshotting? how many snapshots are lost in the remote machines? 2008-01-10 12:49 we dont replicate the scheduled snapshots currently, that is an open issue though 2008-01-10 12:49 which we are probably going to add soon 2008-01-10 12:49 but was that the question? 2008-01-10 12:49 he wants to see hourly/daily snapshots on the downstream 2008-01-10 12:49 i think 2008-01-10 12:49 pgquiles, is that right? 2008-01-10 12:50 yes 2008-01-10 12:51 flips: reading the issue tracker I came to think snapshots were replicated but only most recent snapshots were kept 2008-01-10 12:51 we take a new snapshot for replication at the moment 2008-01-10 12:54 pgquiles, could you explain to me why it is important to replicate the exact hourly snapshots,? 2008-01-10 12:54 I _think_ I see what you really want 2008-01-10 12:54 which is that the downstream should have a reasonable set of snapshots going back some time, not just the most recent replicated snapshots 2008-01-10 12:55 if replication cycle is short, the way we do it now, downstream would not go very far back in time 2008-01-10 12:55 flips: actually, I'd prefer exactly the same snapshots 2008-01-10 12:55 could you explain your reasoning? 2008-01-10 12:55 yes 2008-01-10 12:55 I have 4 servers 2008-01-10 12:56 two at the central office, 1 at a branch office, and 1 at anothe branch office 2008-01-10 12:56 I'd like everybody to see exactly the same data, regardless where they are 2008-01-10 12:57 what if your churn on the volume creates a delta that takes more than an hour to send? 2008-01-10 12:57 if someone deletes a file in office 1 and is available in office 2 but not in office 3 because of differents snapshots, it's a bit of a mess 2008-01-10 12:57 flips: I had not considered that, I didn't know deltas should need less than 1 hour to transfer 2008-01-10 12:58 say I am doing daily snapshots 2008-01-10 12:58 why the 1 hour limit? 2008-01-10 12:58 that would be if you were doing hourlies 2008-01-10 12:58 I'd have the whole 24 hours to transfer the delta 2008-01-10 12:58 flips: ah, ok 2008-01-10 12:58 I thought I was missing something :-) 2008-01-10 12:58 but we still need to handle the case where replication takes longer than the schedule interval 2008-01-10 12:59 well, in that case I think the best approach would be to keep transfering the delta you are transfering and cancel snapshots (up to a reasonable number of cancels) 2008-01-10 12:59 say I'm doing hourlies and my delta takes 2h 20 min to transfer 2008-01-10 12:59 the next 2 deltas would be cancelled 2008-01-10 13:00 then when time for the 3rd delta arrives, there goes the data for every change since the last successfully transferred delta 2008-01-10 13:00 a reasonable approach for some situations 2008-01-10 13:01 we have concentrated on "as fresh as possible" so far 2008-01-10 13:01 would you be willing to write a cron script to accomplish what you want? 2008-01-10 13:01 I would even have an special case: the first replication is likely to exceed the time allowed for delta transmission 2008-01-10 13:02 flips: I know scripting and cron but I have not started with zumastor yet 2008-01-10 13:02 we have a "zumastor replication " command meant to be used from a script 2008-01-10 13:02 I hope to finish a webapp tomorrow or on Monday, probably next week I'll start testing zumastor 2008-01-10 13:02 ok 2008-01-10 13:03 we have also discussed the idea of named snapshots 2008-01-10 13:03 that'd be nice, too 2008-01-10 13:03 so maybe you could do "zumastor replicate " 2008-01-10 13:03 flips: hi 2008-01-10 13:03 hi ejt 2008-01-10 13:03 flips: I'd call it tagged snapshot, stealing svn naming :-P 2008-01-10 13:03 ejt is joe thornber, inventor device mapper and thus of modern lvm on linux 2008-01-10 13:04 ACTION bows 2008-01-10 13:05 pgquiles, so one would "tag a golden snapshot" 2008-01-10 13:05 flips: yeah, why not. It's the same we do with svn or cvs. 2008-01-10 13:05 in a sense, the snapshot number is already a tag, in fact that is what it is called internal to ddsnap 2008-01-10 13:06 the same way the revision number is a "tag" to subversion 2008-01-10 13:06 is it? 2008-01-10 13:06 I don't know internally, but for users is 2008-01-10 13:07 anyway, everybody likes the idea of having strings to name snapshots, as an alternative to the numbers 2008-01-10 13:07 what's the difference between "checkout zumastor revision 1264 from trunk" and "checkout branch 0.5"? 2008-01-10 13:07 yes 2008-01-10 13:07 s/branch 0.5/tag 0.5 2008-01-10 13:08 now if we could implement filters or something like that, we could even have "branches" (views of the data) 2008-01-10 13:08 :-) 2008-01-10 13:08 re svn, there is a difference. Tags in svn are just different directories in one big tree, of which trunk is just one branch 2008-01-10 13:08 revision numbers apply to the big tree including tags and branches, as a whole 2008-01-10 13:09 ejt, is the baby keeping you busy? 2008-01-10 13:11 screaming as I type 2008-01-10 13:11 pgquiles, ok, everybody likes being able to name snapshots, suppose we let you write "zumastor snapshot " and then "zumastor replicate " 2008-01-10 13:13 once they get to the first target, they then must be replicated exactly to further downstream levels, if you have multiple levels of replication 2008-01-10 13:15 do the "nodes" in that mesh know the structure or only where they replicate from and where they replicate to? 2008-01-10 13:18 only from and to 2008-01-10 13:18 because that way a complex mesh can be represented without a complex database 2008-01-10 13:19 somebody can then write a higher level application that goes out and constructs the whole mesh by contacted each node in turn 2008-01-10 13:21 ok 2008-01-10 13:22 pgquiles, for now I am thinking that named snapshots and your exact replication case are tied together 2008-01-10 13:23 why? 2008-01-10 13:23 naming a snapshot somehow makes it seem special 2008-01-10 13:23 and therefore worth replicating 2008-01-10 13:24 I think it looks special only to humans 2008-01-10 13:24 and to scripts, because it has a name 2008-01-10 13:25 as long as a snapshot is kept, zumastor should not care if you are replicating a number or a name 2008-01-10 13:25 "kept" meaning "it has not been deleted yet" 2008-01-10 13:25 typically, people want the replication target to have the freshest possible data 2008-01-10 13:25 whic is why we take a snapshot just before replicating 2008-01-10 13:26 I think that's a different issue 2008-01-10 13:26 right, but we need to be able to accomodate both 2008-01-10 13:27 for example, I may want to have daily snapshots but name only once a week 2008-01-10 13:27 are snapshots still number-based or are they date-based? 2008-01-10 13:27 number-based 2008-01-10 13:27 recycling numbers? 2008-01-10 13:28 (I'm sorry I don't know any of the details, I've been quite busy and unable to read the howto) 2008-01-10 13:28 if they wrap past 2**32,yes 2008-01-10 13:28 which would take about 25 years at one snapshot per second 2008-01-10 13:28 I guess that means "no" for replication purposes :-) 2008-01-10 13:28 I would agree 2008-01-10 13:29 say B and C replicate from A 2008-01-10 13:29 "A" is taking hourly snapshots 2008-01-10 13:29 B and C replicate from A every 6 hours 2008-01-10 13:30 you'd need to replicate revisions N, N+1, ..., N+5 2008-01-10 13:30 regardless they have a name or not, you just ask A for revision number N 2008-01-10 13:30 or N+1 2008-01-10 13:31 that a nice example 2008-01-10 13:31 replicating the exact hourly snapshot makes the most sense then 2008-01-10 13:31 so there is no question we want to support doing exactly what you want there 2008-01-10 13:32 but we need to come up with a sensible command model that also accomodates the "replicate as fresh as possible" case we support now 2008-01-10 13:32 and also supports combinations of the two 2008-01-10 13:32 indeed 2008-01-10 13:33 and it'd be nice if we could avoid sending two deltas with the same information ("trunk" and "N+5") 2008-01-10 13:33 sending an empty delta is practically free 2008-01-10 13:33 I don't mean that 2008-01-10 13:33 I mean sending the same data twice 2008-01-10 13:34 you would be sending an empty delta if you replicate trunk after N+5 2008-01-10 13:34 the same data never gets sent twice 2008-01-10 13:34 but if you are doing trunk at the same time time you are doing N+5, or while you are doing, say, N+3, you'd be sending data twice 2008-01-10 13:34 ok 2008-01-10 13:34 each delta is always relative to the one before it 2008-01-10 13:34 mmm 2008-01-10 13:35 does that mean you can only replicate trunk after you have replicated N+5? 2008-01-10 13:35 this is what zumastor does, at the ddsnap level you can do more complex things if you want 2008-01-10 13:35 atm zumastor is complex enough to keep me busy :-) 2008-01-10 13:37 trunk can be replicated before or after n+5, in the first case, the delta trunk -> n+5 is sent, in the second case the delta n+5 -> trunk is sent, if I have the correct understanding of what you mean by trunk 2008-01-10 13:37 trunk == freshest 2008-01-10 13:37 what happens if N+5 if previous to trunk 2008-01-10 13:37 forget that 2008-01-10 13:38 it's just impossible :-) 2008-01-10 13:38 ok, that's what I understood by trunk 2008-01-10 13:39 or not? 2008-01-10 13:39 impossible to send redundant delta, yes 2008-01-10 13:39 mmm I need to think about that 2008-01-10 13:39 that is nice, don't you think? 2008-01-10 13:39 that's great 2008-01-10 13:40 so, in my mind "trunk" means "data we have stored but we may have not taken a snapshot yet" 2008-01-10 13:41 meaning a delta trunk -> n+5 cannot exist 2008-01-10 13:41 but a delta n+5 -> trunk can exist 2008-01-10 13:41 s/can/may 2008-01-10 13:41 can deltas be "reverse"? 2008-01-10 13:42 they can 2008-01-10 13:42 you can do some really fancy things by playing with that 2008-01-10 13:42 ok, then delta trunk -> n+5 can exist 2008-01-10 13:42 I have yet to understand all the magic in zumastor 2008-01-10 13:42 actually, "trunk" is always a snapshot, what makes it special is, it is the most recent snapshot 2008-01-10 13:43 true, you always take a snapshot before replicating 2008-01-10 13:43 if trunk were not a snapshot, but the actual current copy of the upstream volume, that would be a mirror, not a snapshot 2008-01-10 13:43 I'm still thinking to simplistically 2008-01-10 13:44 we do want to have mirrors too, that is the "ddraid" code 2008-01-10 13:44 in the near future, you mean? 2008-01-10 13:44 it will take some time to bring the ddraid code to a usable form 2008-01-10 13:45 currently it has just enough implemented to get benchmarks out of it 2008-01-10 13:46 flips: do you have much to do with the LVM guys anymore ? 2008-01-10 13:46 ejt, I have a general agreement with alasdair to show some code 2008-01-10 13:47 and I keep in contact with Heinz 2008-01-10 13:48 ejt, it is about time to form a committe on the future of lvm I think, consisting of the four usual suspects name here 2008-01-10 13:49 everybody else welcome of course, but the problem is, for most people their eyes just glaze over when the subject comes up 2008-01-10 13:51 hmm, I see the number one google hit on "cluster raid" is my old red hat source code 2008-01-10 13:52 even funnier, ddsnap is only hit number two for "cluster snapshot" 2008-01-10 13:54 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-10 13:54 hi jiayingz 2008-01-10 13:55 jiayingz, ejt is joe thornber, inventor of device mapper 2008-01-10 13:55 ejt, jiayingz is our resident PhD storage hacker 2008-01-10 13:57 hi jiayingz 2008-01-10 13:57 are you all google here ? 2008-01-10 13:57 pgquiles, lets see if we can come up with a proposal for a zumastor extension that handles your use case nicely, we should probably take it to the mailing list now 2008-01-10 13:58 ejt, about half and half 2008-01-10 13:58 jayingz, shapor, flipz and nataliep_ are googlers, zumalog is a bot 2008-01-10 13:58 flips: agreed 2008-01-10 13:59 y, I guessed about zumalog ;) 2008-01-10 13:59 in some sense, zumalog is not a googler, particularly as it lives on shapor's server 2008-01-10 14:01 ejt, I would like to move all of devmapper into the normal block layer so that somebody could for example snapshot hda without any preplanning, what think you? 2008-01-10 14:04 I think it's been around long enough that people would accept that 2008-01-10 14:04 you'd need Jens to agree, which I suspect he would 2008-01-10 14:05 my main concern is userland tools 2008-01-10 14:05 they just don't seem to have progressed 2008-01-10 14:06 I wrote a pretty nice alternative interface the devmapper that is much easier to use from a C program 2008-01-10 14:06 I'd be interested in seeing that 2008-01-10 14:06 the userland<->kernel interface was always ugly 2008-01-10 14:07 ok, I'll email it privately, I don't mind posting it to dm-dev except the patch needs to be freshened 2008-01-10 14:08 flips: I don't read any mailing lists ATM, you'll have to CC me 2008-01-10 14:08 same with me, I'm subscribed but only read what I'm cc'd on 2008-01-10 14:11 -!- cbsmith(~xman@adsl-71-133-80-65.dsl.irvnca.pacbell.net) has joined #zumastor 2008-01-10 14:12 hi ejt 2008-01-10 14:43 -!- ejt_(~ejt@85-211-138-190.dyn.gotadsl.co.uk) has joined #zumastor 2008-01-10 14:45 wb ejt_ 2008-01-10 14:46 hmmm 2008-01-10 14:50 ACTION patiently waits for the kernel to download.... 2008-01-10 15:04 ACTION will reappear shortly as flipz 2008-01-10 15:05 ejt_ I guess it is midnight over there 2008-01-10 15:07 just gone 11pm, Alice has colic, so carrying her up and down 2008-01-10 15:07 just one thought: re your longstanding "begin cruft", I can see removing the need to clone every incoming bio by providing just a little bit more memory in incoming bios, which is a fairly straightforward allocation effort along the lines of what was done for inodes in 2.6.early 2008-01-10 15:07 err, 2.5.late I mean 2008-01-10 15:08 ajt_, good luck with that colic, been there 2008-01-10 15:09 I've always felt that the kernel shouldn't accept an io unless it has enough memory to process it 2008-01-10 15:09 ie. it should block on the io call until it does 2008-01-10 15:10 at the moment it can bite of more than it can chew 2008-01-10 15:10 (block the userland process that is) 2008-01-10 15:12 http://www.tgdaily.com/content/view/34065/135/ 2008-01-10 15:14 ejt_, I've already implemented the part where the submit_bio blocks when memory is not guaranteed, this is now we prevent deadlock in ddsnap 2008-01-10 15:23 shapor, 832 GB to be delivered in 3G 2008 as I recall 2008-01-10 15:24 not shipping yet. Compared to violin, not raided, hot swappable or shareable between machines 2008-01-10 15:25 likely will cost over $10k, which will give it a certain niche 2008-01-10 15:33 flips: Aren't those the guys I asked you about last month? 2008-01-10 15:34 new entrant 2008-01-10 15:34 oh wait, no 2008-01-10 15:34 actually _I_ was talking about a new entrant 2008-01-10 15:34 whose name I forget 2008-01-10 15:34 "lets start an SSD company" seems to be a propular theme these days 2008-01-10 15:35 http://www.engadget.com/2008/01/04/bitmicro-introduces-832gb-ssd-for-ces-2008/ <- bitmicro 2008-01-10 15:40 flips: yes, I've noticed that. Some kind of key threshold has been passed. 2008-01-10 15:40 I think someone should make a 3.5" form factor drive that holds some large number of SD cards, since they are so cheap now 2008-01-10 15:41 willn: Gigabyte has kind of done that with RAM with their whole i-RAM lineup. 2008-01-10 15:42 willn: But I am kind of disappointd that they don't a) allow one to install more RAM and b) haven't upgraded to at least a SATA-II interface. 2008-01-10 15:43 willn: There is such a huge opportunity there to be a better RAID controller with software RAID + an i-RAM style system that it annoys me that it isn't quite possible. 2008-01-10 15:43 Yea. 2008-01-10 15:45 I can only suppose that gigabyte didn't sell enough iram cards to justify a design rev 2008-01-10 15:45 if true, it means people are doing more wanting than buying 2008-01-10 15:46 Those cards reminded me of the ram doubler cards I used to stuff in my Pentium IIs many years ago 2008-01-10 15:47 flips: I also think that you need software to really take advantage of it properly. The main crowd this got targeted towards were the gamers, who wanted fast load times for their games. The real market should be servers that want really high TPC and process too much data for the puny RAM caches on most RAID controllers to help. 2008-01-10 15:48 software indeed 2008-01-10 15:48 maybe we better make some 2008-01-10 15:48 ACTION looks at zumastor... ;-) 2008-01-10 15:49 I think the necessary hackery would probably need to live above the block device layer, unless the RAM disk was purely used as a cache. 2008-01-10 15:49 flips: thats sata attach, the fusion io is pci-e 2008-01-10 15:49 shapor: Yeah, FusionIO is much closer to Violin in design. 2008-01-10 15:51 me really heads for the office this time 2008-01-10 15:51 /me even 2008-01-10 17:03 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-10 18:55 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-10 20:33 shapor: you around? 2008-01-10 20:34 shapor: I should have an image for you soon. Somehow vmware corrupted my original image *and* hung my machine, but after I put my son to sleep I should have a nice new image. 2008-01-10 23:22 shapor: ping? 2008-01-10 23:58 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Fri Jan 11 00:00:01 PST 2008 2008-01-11 00:00 Hmmm... somehow my image is too big now. Weird. 2008-01-11 00:05 230MB vs the old image which is 191... I've obviously forgotten to do something. 2008-01-11 00:05 ACTION smacks self 2008-01-11 00:05 Ugh. I forgot to clean up the kernel directory. 2008-01-11 00:36 -!- ejt(~ejt@85-211-138-190.dyn.gotadsl.co.uk) has joined #zumastor 2008-01-11 01:02 morning ejt 2008-01-11 01:02 morning 2008-01-11 01:05 baby sleeping? 2008-01-11 01:06 y, I went to bed at 1am, I gather Kath was up alot in the night though 2008-01-11 01:06 separate beds was the solution for us 2008-01-11 01:07 well, anyway, may I mumble a bit about thoughts on dm improvements? 2008-01-11 01:07 sure 2008-01-11 01:08 I'm sure you know what I'm talking about when I mention clone+hook 2008-01-11 01:08 hooking endio is really a pain 2008-01-11 01:08 y 2008-01-11 01:09 if there was a some working space in each bio, it can be done nicely 2008-01-11 01:09 so for example, each layer of a stack could own x bytes of private space in the bio 2008-01-11 01:09 this layout would only change if the stack was reconfigured 2008-01-11 01:10 just doing this at the top level, where dm hooks the endio and thus has to clone would already be a good thing 2008-01-11 01:11 so you're talking about variable sized bios ? 2008-01-11 01:11 yes 2008-01-11 01:11 ie. the device specifies how much space it needs to process 2008-01-11 01:12 or maybe have that memory in the biovec, and have the bio be a header on biovec 2008-01-11 01:12 right 2008-01-11 01:12 that _sounds_ like the sort of thing I wanted 4 years ago when I was last thinking about this 2008-01-11 01:13 it's not an easy problem 2008-01-11 01:13 as ever I think we'd have to try coding it up and see - I suspect it's better to keep the bios a fixed size 2008-01-11 01:13 and hook a chunk of memory onto them 2008-01-11 01:13 stacking devices is getting more important all the time, and some of the new block devices are very fast, so reduction allocations in the handler path actually matters 2008-01-11 01:14 each bio already gets a variable sized biovec 2008-01-11 01:14 obvious thing to do is make the bio a header on the biovec 2008-01-11 01:14 so they no longer come from a slab ? 2008-01-11 01:14 agreed 2008-01-11 01:14 they would have to come from an array of slabs, which they already do 2008-01-11 01:15 similar to kmalloc, but finer granularity 2008-01-11 01:15 memory allocation deadlock issues ahve been dealt with by the way 2008-01-11 01:15 that was hard 2008-01-11 01:16 I gather bios go back to the start of the queue when they get remapped 2008-01-11 01:16 took several months to work around the ineffective attempts that had been added over the last few years to deal with writeout deadlocks 2008-01-11 01:16 agk didn't sound too happy about it 2008-01-11 01:16 go back to the start of the queue? 2008-01-11 01:17 I need to grab the kernel source at set up yet another VM 2008-01-11 01:17 we have a nice little uml environment you can download and run 2008-01-11 01:17 on zumastor.org 2008-01-11 01:18 oh, you mean remapped to a different place on the same device, hmm 2008-01-11 01:18 I don't know what you mean 2008-01-11 01:18 there is no queue in front of a dm device 2008-01-11 01:19 let me grab code and have a read, rather than repeating stuff agk has told me 2008-01-11 01:19 lxr.linux.no is pretty good 2008-01-11 01:19 running lxr 0.9, interface is a little funky, but it works and has a good grep 2008-01-11 01:20 time for me to sleep 2008-01-11 01:20 ok, talk this evening then maybe ? 2008-01-11 01:21 sure 2008-01-11 09:24 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-11 09:24 -!- charlesnw(~charles@ses.siderean.com) has left #zumastor 2008-01-11 10:14 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-11 10:20 good morning 2008-01-11 10:35 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-11 10:57 Morning folks 2008-01-11 10:57 ACTION heads to ARI 2008-01-11 10:58 -!- willn_(~wan@pinball.ccs.neu.edu) has joined #zumastor 2008-01-11 13:55 The Sun X4500 has a 48x1TB configuration now 2008-01-11 13:55 willn: yup 2008-01-11 13:56 willn: Bets it's using WD GPs? ;-) 2008-01-11 13:56 Last I heard they were only making a 500G model 2008-01-11 13:56 they also have a 250G and 750G model 2008-01-11 13:56 willn: I just watched a presentation on their new product line (literally finished it about 15 minutes ago). ;-) 2008-01-11 13:57 ah 2008-01-11 13:58 I'm somewhat suprised they don't offer a hardware raid version for people running Windows 2008-01-11 14:12 willn: Do people run Windows on the x4500? 2008-01-11 14:12 Yep irc.oftc.net #zumastor log beginning Sat Jan 12 00:00:01 PST 2008 2008-01-12 03:26 -!- pgquiles(~pgquiles@121.Red-81-38-97.dynamicIP.rima-tde.net) has joined #zumastor irc.oftc.net #zumastor log beginning Sun Jan 13 00:00:01 PST 2008 2008-01-13 00:30 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-13 22:25 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Jan 14 00:00:01 PST 2008 2008-01-14 00:49 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-14 07:56 ACTION blinks 2008-01-14 08:10 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-14 09:29 -!- nataliep_(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-14 11:01 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-14 11:01 shapor: ping 2008-01-14 11:26 Tim_vimm: pong 2008-01-14 11:26 hey 2008-01-14 11:26 you in MV yet? 2008-01-14 11:27 yep 2008-01-14 11:27 got in yesterday 2008-01-14 11:28 need to work out some details...when, where, parking, security 2008-01-14 11:28 when would be a good time to talk? 2008-01-14 11:28 oh yeah, drivers too 2008-01-14 12:19 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-14 13:25 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-14 15:01 -!- flipz(~phlipz@216-239-45-4.google.com) has joined #zumastor 2008-01-14 15:01 good day 2008-01-14 15:01 greetings 2008-01-14 15:45 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-14 19:02 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-14 19:43 -!- flipz(~phlipz@adsl-63-202-13-187.dsl.snfc21.pacbell.net) has joined #zumastor 2008-01-14 20:21 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-14 20:55 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-14 22:18 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-14 23:00 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-14 23:03 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Tue Jan 15 00:00:01 PST 2008 2008-01-15 01:14 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-15 01:18 -!- erwan__taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-15 01:33 -!- erwan__taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-15 07:18 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-15 08:46 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-15 10:26 -!- flipz(~phlipz@adsl-63-202-13-187.dsl.snfc21.pacbell.net) has joined #zumastor 2008-01-15 10:26 good morning 2008-01-15 10:32 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-15 10:44 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-15 11:13 ACTION hmms https://wiki.ubuntu.com/MultiMachineTestingInfraSpec 2008-01-15 11:20 hi daniel 2008-01-15 12:02 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-15 13:18 -!- flipz(~phlipz@216-239-45-4.google.com) has joined #zumastor 2008-01-15 15:21 -!- flipz(~phlipz@216-239-45-4.google.com) has joined #zumastor 2008-01-15 16:13 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-15 16:55 -!- flipz(~phlipz@216-239-45-4.google.com) has joined #zumastor 2008-01-15 17:03 -!- mfedyk(~mfedyk@adsl-63-194-240-129.dsl.lsan03.pacbell.net) has joined #zumastor 2008-01-15 17:03 hello 2008-01-15 17:04 I'm reading through the zumastor whitepaper right now 2008-01-15 17:06 mfedyk: welcome 2008-01-15 17:06 and I see several improvements compared to drbd, but wonder why that code base wasn't branched instead of starting over 2008-01-15 17:06 ie, it looks like zumastor can snapshot a snapshot 2008-01-15 17:06 which isn't available functionality in DM 2008-01-15 17:07 but is DM so broken that it needs to be replaced? 2008-01-15 17:07 shapor: hi 2008-01-15 17:07 zumastor is quite different from drbd and lvm snapshots 2008-01-15 17:07 lvm snapshots keep a copy of the data for every snapshot 2008-01-15 17:07 is it based on csnap? 2008-01-15 17:07 it is csnap 2008-01-15 17:08 which became ddsnap 2008-01-15 17:08 which is what we've been working on mostly 2008-01-15 17:09 yes, and because DM keeps data in each snapshot instead of keeping track of the bitmaps and how they overlap in the snapshots it becomes that much more difficult to snapshot a snapshot 2008-01-15 17:09 so when you want to take several snapshots, space and performance wise lvm snapshots are useless 2008-01-15 17:11 hi mfedyk 2008-01-15 17:11 ACTION <- daniel phillips 2008-01-15 17:11 is the overall design of DM conducive to keeping track of PEs like ddsnap does? 2008-01-15 17:11 hey flipz 2008-01-15 17:12 I see your name in this and I know there has to be something I'm missing 2008-01-15 17:12 mfedyk, there is very little that could be usefully carried over from the existing lvm snapshot 2008-01-15 17:12 little if anything 2008-01-15 17:13 how about carried over to DM? 2008-01-15 17:13 but dm-ddsnap is a device mapper target 2008-01-15 17:13 or would that just be another painful on disk format change? 2008-01-15 17:13 ahhhh 2008-01-15 17:13 so it works within the DM framework 2008-01-15 17:13 nice 2008-01-15 17:14 yes 2008-01-15 17:14 snapshotting a snapshot is not completely easy with ddsnap either 2008-01-15 17:14 not implemented yet 2008-01-15 17:14 in principle, it requires walking the btree and duplicating each share bit for a given snapshot 2008-01-15 17:15 not hard, but it is not instantaneous 2008-01-15 17:15 so I could set CONFIG_DM_SNAPSHOT=n and replace it with ddsnap 2008-01-15 17:15 it's not quite that compatible 2008-01-15 17:15 oh 2008-01-15 17:15 yes you can 2008-01-15 17:16 but lvm does not know about how to make ddsnap take a snapshot 2008-01-15 17:16 you have to give the command using the ddsnap utility 2008-01-15 17:16 why is that? 2008-01-15 17:16 easier to specify replication? 2008-01-15 17:17 or is it just a userspace tools issue? 2008-01-15 17:17 lvm has no concept of replication at all 2008-01-15 17:17 so you use the zumastor script instead 2008-01-15 17:19 is there any functionality provided in the stock dm-snapshot that isn't covered by ddsnap? 2008-01-15 17:21 so ddsnap goes in and the need for drbd goes away? 2008-01-15 17:22 from the whitepaper, it looks like you can mount a read-only filesystem in a replication target. 2008-01-15 17:22 does that mean it is pagecache coherent? 2008-01-15 17:31 no, its asyncronous replication 2008-01-15 17:32 you essentially just have read only access to some snapshot which has been transmitted 2008-01-15 17:36 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-16 08:07 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2008-01-16 08:55 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-16 09:02 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-16 09:09 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-16 09:23 good morning 2008-01-16 09:43 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-16 10:02 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2008-01-16 10:12 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2008-01-16 10:20 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2008-01-16 11:59 morning 2008-01-16 12:12 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-16 12:29 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-16 14:03 -!- flipz(~phlipz@216.239.45.19) has joined #zumastor 2008-01-16 17:10 -!- flipz(~phlipz@216.239.45.19) has joined #zumastor 2008-01-16 19:42 Procient with Active Directory, Apache, MySQL, Samba, Sun Enterprise System 2008-01-16 19:43 erm, ignore that. 2008-01-16 19:53 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-16 23:00 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-16 23:18 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Thu Jan 17 00:00:01 PST 2008 2008-01-17 00:04 -!- flipz(~phlipz@adsl-63-202-13-187.dsl.snfc21.pacbell.net) has joined #zumastor 2008-01-17 07:32 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-17 09:06 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has left #zumastor 2008-01-17 09:06 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-17 09:07 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-17 11:17 -!- flipz(~phlipz@216-239-45-4.google.com) has joined #zumastor 2008-01-17 11:22 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-17 11:28 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has left #zumastor 2008-01-17 12:40 -!- pgquiles(~pgquiles@81.202.65.108) has joined #zumastor 2008-01-17 13:50 -!- flipz(~phlipz@216-239-45-4.google.com) has joined #zumastor 2008-01-17 13:56 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-17 14:12 jiayingz: ping 2008-01-17 14:18 hi pgquiles 2008-01-17 14:18 hi pgquiles 2008-01-17 14:18 hi jiayingz flipz 2008-01-17 14:18 jiayingz: nothing :-) 2008-01-17 14:19 jiayingz: I've just replied to your e-mail on snapshot replication 2008-01-17 14:20 I saw that 2008-01-17 14:20 thanks for the reply 2008-01-17 14:20 I'm working on a nice web 2.0 gui for our servers, including zumastor support, btw :-) 2008-01-17 14:20 that will be great! 2008-01-17 14:20 it's still very basic and buggy but it looks nice 2008-01-17 14:21 so do you need that feature you described for the web servers to work? 2008-01-17 14:21 no, not for the webservers 2008-01-17 14:21 each server is essentially a NAS + qemu'ed virtual machines 2008-01-17 14:21 and I want to replicate everything 2008-01-17 14:21 i c 2008-01-17 14:22 let us know if you need any feature from zumastor 2008-01-17 14:22 including the admin tool, which will know about the nodes in the mesh and show info about everything 2008-01-17 14:22 sure 2008-01-17 14:22 right now I only need more time :-) 2008-01-17 14:23 same here. A lot of features should be added, just need more time :) 2008-01-17 14:28 http://img444.imageshack.us/my.php?image=pfsilt4.png 2008-01-17 14:33 the picture of the web interface? looks nice 2008-01-17 15:09 -!- flipz(~phlipz@216.239.45.19) has joined #zumastor 2008-01-17 15:54 http://www.fpsbrain.com/?switchLanguage=english 2008-01-17 15:55 :-) 2008-01-17 16:46 fmayhar: lol 2008-01-17 17:55 -!- flipz(~phlipz@adsl-63-202-13-187.dsl.snfc21.pacbell.net) has joined #zumastor 2008-01-17 19:06 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-17 21:47 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-17 22:29 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-17 22:30 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Fri Jan 18 00:00:01 PST 2008 2008-01-18 07:11 -!- charlesnw(~charles@cpe-75-84-87-237.socal.res.rr.com) has joined #zumastor 2008-01-18 08:41 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-18 10:34 -!- flipz(~phlipz@216-239-45-4.google.com) has joined #zumastor 2008-01-18 11:41 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-18 11:42 Howdy... any flipz here? 2008-01-18 11:42 hi dkegel 2008-01-18 11:42 what up? 2008-01-18 11:43 Think you'll have a chance to review Jiaying's shrink patches soon? 2008-01-18 11:43 yes 2008-01-18 11:43 subject line? 2008-01-18 11:44 "origin/snapshot resizing". She has two patches in there, I'll ask her to repost for clarity. 2008-01-18 11:45 I'd like to get all 0.6 non-regression patches in by Monday, per the schedule I proposed on the 11th. 2008-01-18 11:51 Whoops, Monday's a holiday, guess that's a one-day slip. 2008-01-18 11:57 insta feedback is: there is a generic message reply handler in ddsnap.c, please use it instead of coding the reply handling in-line. If it doesn't work for this application for some reason, then fix the generic handler 2008-01-18 11:57 off two lunch 2008-01-18 14:42 -!- charlesnw(~charles@cpe-75-84-87-237.socal.res.rr.com) has left #zumastor 2008-01-18 15:56 -!- flipz(~phlipz@216.239.45.19) has joined #zumastor 2008-01-18 16:06 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-18 17:52 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sat Jan 19 00:00:01 PST 2008 2008-01-19 01:03 -!- willn(~wan@pinball.ccs.neu.edu) has joined #zumastor 2008-01-19 01:04 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-19 01:10 -!- flipz(~phlipz@phunq.net) has joined #zumastor 2008-01-19 05:12 -!- pgquiles(~pgquiles@157.Red-81-44-62.dynamicIP.rima-tde.net) has joined #zumastor 2008-01-19 12:50 -!- flipz(~phlipz@phunq.net) has joined #zumastor 2008-01-19 20:29 -!- cbsmith(~xman@67.201.162.41) has joined #zumastor 2008-01-19 21:49 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2008-01-19 21:49 -!- willn(~wan@pinball.ccs.neu.edu) has joined #zumastor 2008-01-19 21:49 -!- pgquiles_(~pgquiles@157.red-81-44-62.dynamicip.rima-tde.net) has joined #zumastor 2008-01-19 21:49 -!- cbsmith(~xman@67.201.162.41) has joined #zumastor 2008-01-19 21:49 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-19 21:49 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-19 21:49 -!- nataliep_(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-19 21:49 -!- flips(~phillips@phunq.net) has joined #zumastor 2008-01-19 21:49 -!- juuva(juuva@peili.org) has joined #zumastor irc.oftc.net #zumastor log beginning Sun Jan 20 00:00:01 PST 2008 2008-01-20 09:25 -!- charlesnw(~charles@cpe-75-84-87-237.socal.res.rr.com) has joined #zumastor 2008-01-20 10:28 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 10:56 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor 2008-01-20 11:26 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2008-01-20 11:46 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 12:38 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 15:52 hey 2008-01-20 15:52 time to do bio_push/bio_pop now 2008-01-20 15:52 soon we will make right what is wrong in the block layer 2008-01-20 17:49 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 18:17 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 18:35 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 19:42 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 20:34 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-20 20:52 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor 2008-01-20 22:10 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Jan 21 00:00:01 PST 2008 2008-01-21 01:02 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-21 03:53 -!- pgquiles(~pgquiles@151.Red-213-96-197.staticIP.rima-tde.net) has joined #zumastor 2008-01-21 03:54 -!- pgquiles(~pgquiles@151.Red-213-96-197.staticIP.rima-tde.net) has joined #zumastor 2008-01-21 08:12 -!- pgquiles(~pgquiles@151.Red-213-96-197.staticIP.rima-tde.net) has joined #zumastor 2008-01-21 09:01 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor 2008-01-21 09:14 -!- pgquiles(~pgquiles@151.Red-213-96-197.staticIP.rima-tde.net) has joined #zumastor 2008-01-21 09:40 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-21 10:26 -!- mitchg_(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-21 10:56 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor 2008-01-21 11:14 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor 2008-01-21 11:33 -!- flips(~phillips@phunq.net) has joined #zumastor 2008-01-21 12:17 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor 2008-01-21 14:02 -!- charlesnw(~charles@cpe-75-84-88-252.socal.res.rr.com) has joined #zumastor 2008-01-21 15:37 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-21 19:24 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor irc.oftc.net #zumastor log beginning Tue Jan 22 00:00:01 PST 2008 2008-01-22 00:01 -!- willn(~wan@pinball.ccs.neu.edu) has joined #zumastor 2008-01-22 00:37 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-22 00:40 -!- erwan_taf(~erwan@LAubervilliers-151-13-63-69.w217-128.abo.wanadoo.fr) has joined #zumastor 2008-01-22 01:00 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-22 01:43 -!- shapor_(~shapor@yzf.shapor.com) has joined #zumastor 2008-01-22 01:47 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-22 01:53 -!- flips(~phillips@phunq.net) has joined #zumastor 2008-01-22 03:41 ACTION most likely getting up around noon... later today 2008-01-22 07:33 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-22 09:02 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-22 09:38 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-22 09:40 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has left #zumastor 2008-01-22 09:40 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-22 09:47 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-22 10:33 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-22 10:44 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-22 10:56 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-22 11:00 good morning 2008-01-22 12:18 howdy all 2008-01-22 13:38 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-22 19:07 ping 2008-01-22 20:50 buy.com has 4G thumb drives for 6$, almost cheap enough to build an array irc.oftc.net #zumastor log beginning Wed Jan 23 00:00:01 PST 2008 2008-01-23 00:29 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-23 07:56 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-23 09:45 -!- charlesnw(~charles@dctm.siderean.com) has joined #zumastor 2008-01-23 10:59 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-23 11:25 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-23 13:20 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-23 14:08 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-23 14:47 jiaying: ping 2008-01-23 15:32 -!- charlesnw(~charles@dctm.siderean.com) has left #zumastor 2008-01-23 15:49 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-23 16:55 Tim_vimm: ping 2008-01-23 16:55 pong 2008-01-23 16:56 We're looking for Maxim to help us sort through some setup issues, the connection info is irc.oftc.net #zumastor 2008-01-23 16:56 think Maxim just bolted for the night 2008-01-23 16:57 I've got Kyle standing by 2008-01-23 16:57 What communication medium is best? 2008-01-23 16:57 how complex is the question. 2008-01-23 16:57 the Tim_vimm switch is here 2008-01-23 16:58 how about a landline 2008-01-23 16:58 We've got everything connected as best we can negotiate, but we're not getting the host interface or Violin box to show up in our lspci output 2008-01-23 16:58 Do we need to load a driver prior to that happening? 2008-01-23 16:58 as far as I know, yes 2008-01-23 16:59 do you have the driver? 2008-01-23 16:59 There are three ports on the back of the VM1010. Is the cable plugged into the one with the GREEN circular light? 2008-01-23 17:00 who's googlechat account should Kyle hit you on? 2008-01-23 17:00 willn@google.com 2008-01-23 17:01 he'll be on irc in a moment 2008-01-23 17:01 -!- kfortin(~chatzilla@smtp.violintech.net) has joined #zumastor 2008-01-23 17:01 Ok, thanks 2008-01-23 17:02 kfortin: Howdy 2008-01-23 17:02 hi all 2008-01-23 17:02 ki kfortin... 2008-01-23 17:02 Does the One Stop Systems interface show up with a obvious vendor id? 2008-01-23 17:02 on the back of the vm1010 where the PCIE cable is plugged in, do you see a green light? 2008-01-23 17:03 Let me double check 2008-01-23 17:03 should be solid green 2008-01-23 17:03 or blinking if no link to the host 2008-01-23 17:03 x8b port 2008-01-23 17:04 lspci will an id of 0x1a73 2008-01-23 17:04 with latest pciutils you can update-pciids and our vendor/model ID will get recognized. 2008-01-23 17:05 ok, we just swapped the connection, now we're getting the blinking light 2008-01-23 17:05 0c:00.0 Memory controller: Violin Memory, Inc Mozart [Memory Appliance 1010] (rev 02) 2008-01-23 17:05 (whereas before we had nothing out of the lefthand side pci-e connection) 2008-01-23 17:06 so the cable is plugged to the port with the blinking light? 2008-01-23 17:06 Now it is 2008-01-23 17:06 ok, so now the host needs to get rebooted for the APCI2.0 BIOS to discover it. 2008-01-23 17:06 Just rebooted. 2008-01-23 17:06 immediately when the BIOS starts, the light will go solid green 2008-01-23 17:08 The VM1010 is the only device I should expect in lspci output, no? 2008-01-23 17:08 you'll see a lot of PCI devices 2008-01-23 17:09 you can use this trick: "lspci -d 1a73: " 2008-01-23 17:09 which only shows any Violin devices 2008-01-23 17:09 rather, I shouldn't see the PCI-E interface card as well 2008-01-23 17:10 the HIB card is a bridge repeater 2008-01-23 17:10 Only the Violin device is seen 2008-01-23 17:12 An interesting thing to look for performance is to do "lspci -d 1a73: -vvv" and check MaxPayload as well as Link speed/width 2008-01-23 17:12 Should see: Link: Speed 2.5Gb/s, Width x8 2008-01-23 17:13 and "Device: MaxPayload 256 bytes, MaxReadReq 512 bytes" or better 2008-01-23 17:13 ah 2008-01-23 17:13 depending on the BIOS support for > 128 byte payloads 2008-01-23 17:15 Is it noted in the manual the PCI-E connectivity preference of port B? 2008-01-23 17:15 I'm pretty sure it is. 2008-01-23 17:16 I notice in the device diagram only B is labeled as being an interface port, we all missed that when racking :) 2008-01-23 17:17 Sorry for the confusion. 2008-01-23 17:18 Makes sense to me to choose A, but I'm not the hardware guy. 2008-01-23 17:18 blame the label guy ;) 2008-01-23 17:20 :) 2008-01-23 17:21 Well, I think we're good. We have to get some kernel/gcc versioning sorted out, but we should be alright from this point forward 2008-01-23 17:22 OK. The R1.2 driver was systested against RHEL5 which was a 2.6.18-8 kernel base, but I believe I compile/tested up to 2.6.20. 2008-01-23 17:22 2.6.20 though had issue with PCI MSI support and needs a "nomsi=1" module parm. 2008-01-23 17:23 It definitely won't compile with 2.6.23.x base as the API changed after R1.2 released. 2008-01-23 17:23 ok 2008-01-23 17:24 -!- phillips_(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-23 17:25 OK, so if you have any more questions, email support@violin-memory.com 2008-01-23 17:25 phillips_:howdy 2008-01-23 17:25 hi tim 2008-01-23 17:25 kfortin: Would 2.6.18-8 be the best bet? 2008-01-23 17:25 willn, what is the exact kernel version the driver wants? 2008-01-23 17:26 yes, for that version of driver. 2008-01-23 17:26 flipz: ^ 2008-01-23 17:26 :) 2008-01-23 17:26 Well the redhat 2.6.18 is not really 2.6.18 2008-01-23 17:26 oh, sweet 2008-01-23 17:27 they backport lots of patches 2008-01-23 17:27 we will try it anyway 2008-01-23 17:27 I've even seen a 2.6.23 patch come in the latest 2.6.18 (minor update) 2008-01-23 17:31 building now 2008-01-23 17:32 from redhat kernel source rpm? or just a 2.6.18 vanilla? 2008-01-23 17:32 mainline 2.6.18.something 2008-01-23 17:32 building silly number of modules, otherwise I would be done by now 2008-01-23 17:33 no problem 2008-01-23 17:33 I think I might stop it and build with make defconfig 2008-01-23 17:33 would be nice to know if that works 2008-01-23 17:33 here we go 2008-01-23 17:33 holding my breath... 2008-01-23 17:34 "time make" :-) 2008-01-23 17:34 fsck? 2008-01-23 17:34 make defconfig is basically linus's own configuration 2008-01-23 17:35 oh, defconfig includes cifs, how odd 2008-01-23 17:35 I usually copy the .config and make oldconfig from a working kernel on the same machine 2008-01-23 17:36 picks the right sata, megaraid_sas, or cciss drivers needed 2008-01-23 17:36 kfortin, exactly what I did, however the working kernel on the same machine includes every module in the world 2008-01-23 17:36 ouch 2008-01-23 17:36 a waste of time to compile all those 2008-01-23 17:36 ccache? 2008-01-23 17:36 in case you have to build again 2008-01-23 17:36 make defconfig :-) 2008-01-23 17:38 :-) 2008-01-23 17:38 even make defconfig is bloated, includes sound and 3D video 2008-01-23 17:38 every geek likes video games, right? 2008-01-23 17:38 playing quake 4 on a server is lots of fun 2008-01-23 17:39 quakeworld was the first app I had up and running on the Itanium HP-UX port 2008-01-23 17:39 real 5m15.928s 2008-01-23 17:39 user 4m54.103s 2008-01-23 17:39 sys 0m26.073s 2008-01-23 17:39 :-) 2008-01-23 17:39 done 2008-01-23 17:42 installed 2008-01-23 17:55 Were you able to install the device driver? 2008-01-23 18:28 kfortin, still mucking with kernel version issues 2008-01-23 18:29 kfortin is putting on his reverse scramjet shoes 2008-01-23 18:31 clicking my heels 3 times 2008-01-23 18:33 make defconfig did not work, the dell raid driver was not recognized 2008-01-23 18:33 so building everything now 2008-01-23 18:33 :p 2008-01-23 18:34 dude, you're getting a dell 2008-01-23 18:34 the kernel also likes to panic and ignore keyboard input in cases like this 2008-01-23 18:34 which is completely idiotic, you can't scroll back to see what went wrong 2008-01-23 18:34 there is a way to change that... don't remember how right now 2008-01-23 20:25 -!- forkyan(~chatzilla@static-host-24-38-199-245.patmedia.net) has joined #zumastor 2008-01-23 20:32 -!- kfortin2(~chatzilla@static-host-24-38-199-245.patmedia.net) has left #zumastor 2008-01-23 21:00 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-23 22:52 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-23 23:26 -!- daniel__(~phlipz@phunq.net) has joined #zumastor irc.oftc.net #zumastor log beginning Thu Jan 24 00:00:01 PST 2008 2008-01-24 00:05 well we are booted with 2.6.18.8 2008-01-24 00:05 mainline 2008-01-24 00:05 have not compiled or tried the violin driver yet 2008-01-24 00:25 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-24 00:57 -!- erwan_taf(~erwan@81.80.43.67) has joined #zumastor 2008-01-24 02:45 -!- flips(~phillips@phunq.net) has joined #zumastor 2008-01-24 02:53 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-24 02:53 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-24 03:00 -!- phillips_(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-24 03:08 -!- juuva(juuva@peili.org) has joined #zumastor 2008-01-24 07:34 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-24 08:21 -!- maximadelman(~maxim@smtp.violintech.net) has joined #zumastor 2008-01-24 08:36 howdy all 2008-01-24 08:36 we tested the 1.2.1 driver last night 2008-01-24 08:36 qualified it in a 12-hour system test run on RHEL5 Server with a 2.6.23.14 kernel.org x86_64 kernel 2008-01-24 08:37 R1.2.1 driver against the VM1010 R1.2 firmware 2008-01-24 08:37 GCC 4.1.1 2008-01-24 08:37 I see flips was a night owl- I've sent the driver to his email 2008-01-24 08:39 if anyone needs it sooner, PM me w/ an email address and I'll forward it along 2008-01-24 08:50 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-24 09:01 Tim_vimm: Thanks 2008-01-24 09:05 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has left #zumastor 2008-01-24 12:04 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-24 12:59 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-24 13:03 -!- maximadelman(~maxim@smtp.violintech.net) has left #zumastor 2008-01-24 13:43 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-24 14:03 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-24 14:33 -!- pgquiles__(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-24 15:13 -!- maximadelman(~maxim@smtp.violintech.net) has joined #zumastor 2008-01-24 15:14 hi willn, this is maxim from violin 2008-01-24 15:14 i just wanted to see if you guys can test the IO performance of your server against our box with our benchmark tool 2008-01-24 15:14 this could give us an idea of how good the server that you are using is 2008-01-24 15:15 we've seen in the past that dell machines are so-so both in terms of raw performance and IO performance against Violin1010, particularly the 2950 2008-01-24 15:16 the test that you'd need to run is compiled as part of the driver and should be in the compilation output directory (e.g. vtms-strad-driver/2.6.23.14/x86_64/) 2008-01-24 15:16 the tool is called perf_test 2008-01-24 15:17 assuming that the driver is loaded and the device is attached, you can run the tool using the following parameters 2008-01-24 15:18 "perf_test -t 8 -b 16384 -o -R -s 64 -d 3" will measure the bandwidth 2008-01-24 15:18 "perf_test -t 16 -b 512 -o -R -s 64 -d 3" will measure the IOPS 2008-01-24 15:19 maximadelman: Alright, we'll give it a try 2008-01-24 15:19 if you can tell me the #s you get, it will give us an idea of the sort of performance we can expect from the server 2008-01-24 15:19 thanks! 2008-01-24 16:00 willn - i am going home now. if you can send results to support@violin-memory.com, that would be great. if you have any Qs - call me, 347.489.5713 anytime before midnight EST 2008-01-24 17:35 -!- fmayhar(~fmayhar@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-24 18:08 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-24 18:15 ./perf_test -t 8 -b 16384 -o -R -s 64 -d 3 -r 1 -w 1 2008-01-24 18:15 phillips_: ^ 2008-01-24 18:22 roughly speaking: about 970 MB/sec raw read, 960 MB/sec raw write on a lowly Dell 1850 2008-01-24 18:22 the moral equivalent of a Volkswagon bug with a jet turbine mounted in the rear window 2008-01-24 18:22 this thread is worthless without pictures ;-) 2008-01-24 18:23 http://www.engadget.com/2006/03/06/jet-powered-vw-bug-takes-to-the-street/ 2008-01-24 18:23 just ask :) 2008-01-24 18:23 tailgate this! 2008-01-24 18:23 You can expect about 1740 MB/s read on an HP DL385 2008-01-24 18:23 and the new bug 2008-01-24 18:23 that's a worthy number 2008-01-24 18:24 but the numbers you got are very respectable 2008-01-24 18:24 try the same test against a mechanical disk and the MB/s will be less than 1 2008-01-24 18:24 that's the only reason we exist 2008-01-24 18:25 make that progress bar go mo faster 2008-01-24 18:26 We are also curious about the IOPS test on that system: perf_test -t 16 -b 512 -o -R -s 64 -d 5 2008-01-24 18:33 You can bump the -s number up to 384 to hit the whole memory range 2008-01-24 18:42 kfortin, I believe you about the 1740 MB/s, we won't attempt to upgrade the machine this time around though 2008-01-24 18:42 the numbers are already impressive 2008-01-24 18:42 and I believe we can trust what you claim about more capable servers 2008-01-24 18:44 kfortin has his scramjet shoes on 2008-01-24 18:44 I already passed that info on to him :) 2008-01-24 18:44 my recollection of the jet powered beetle was slightly wrong, I see it has the engine in the front so the turbine is in the trunk, not the window 2008-01-24 18:44 I doubt it would stay on the road if the turbine was higher 2008-01-24 18:45 what would be really cool is if the roof would open up, and wings sprout out 2008-01-24 18:45 like a lady bug 2008-01-24 18:46 someone versed in the art of CG could make that a virtual reality 2008-01-24 19:06 flipz: where on the zumastor site are the new post you mentioned? 2008-01-24 19:07 nevermind... 2008-01-24 20:04 flipz: well the bandwidth/IOPS will vary greatly. The HP server I mentioned has a PCIE payload of 512 bytes due to the Broadcom HT2100 PCIE bridge chipset. 2008-01-24 20:06 A Dell 2950 with dual quad cores might show 1330MB/s read as it's chipset only allows 256 byte payloads. 2008-01-24 20:07 A Penguin server with 256 byte PCIE payload gets about 1520 MB/s 2008-01-24 20:09 We have sometimes seen, as with the HP machines that they get great bandwidth, but the IOPS might not be as good as say a Penguin 2008-01-24 20:13 Also depends on FSB, DIMM, and CPU speed, and even if the cpuspeed or powersaved daemon(s) are running 2008-01-24 20:17 Also cool is to use the device-mapper in stripe mode and 2 x VM1010's on one HP host showed 3400MB/s read bandwidth 2008-01-24 20:20 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-24 21:56 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Fri Jan 25 00:00:01 PST 2008 2008-01-25 00:09 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-25 00:39 kfortin, thanks for the yummy info. I am indeed interested in which servers suck/don't suck 2008-01-25 06:01 -!- phillips_(~phillips@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-25 06:51 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-25 07:14 -!- charlesnw(~charles@dctm.siderean.com) has joined #zumastor 2008-01-25 07:45 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-25 08:00 -!- maximadelman(~maxim@smtp.violintech.net) has joined #zumastor 2008-01-25 13:01 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-25 13:04 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-25 13:12 -!- maximadelman(~maxim@smtp.violintech.net) has left #zumastor 2008-01-25 18:54 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-25 21:29 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sat Jan 26 00:00:01 PST 2008 2008-01-26 03:54 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-26 07:19 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-26 07:37 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-26 08:29 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-26 12:43 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-26 13:54 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Sun Jan 27 00:00:01 PST 2008 2008-01-27 01:36 -!- eleazar_(~ass@213.231.109.93.dyn.user.ono.com) has joined #zumastor 2008-01-27 04:04 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-27 10:23 -!- kfortin(~chatzilla@smtp.violintech.net) has joined #zumastor 2008-01-27 10:30 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2008-01-27 12:28 -!- zumalog(~zumalog@216.75.12.152) has joined #zumastor 2008-01-27 15:00 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-27 18:46 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Mon Jan 28 00:00:01 PST 2008 2008-01-28 00:25 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-28 00:52 flips: ping 2008-01-28 00:52 hi 2008-01-28 00:54 pgquiles, pong 2008-01-28 00:54 flips: wow, I didn't expect you to answer, it must be really late there :-) 2008-01-28 00:54 only 1 am 2008-01-28 00:54 flips: what patches should I use for kernel 2.6.22.14? 2008-01-28 00:54 the new ones you sent for 2.6.23? 2008-01-28 00:56 if 2.6.22.10 apply cleanly then you are in luck 2008-01-28 00:56 they do 2008-01-28 00:56 yay! that means I don't need to rebuild the kernel 2008-01-28 00:56 :) 2008-01-28 00:57 nobody has tested it of course, let us know how it goes 2008-01-28 00:58 flips: I read the source code for the relevant parts of 2.6.22.10 and 2.6.22.14 and nothing was changed, I assume it works fine 2008-01-28 00:58 I expect so too, but it is always nice to know for sure 2008-01-28 01:00 how long until 0.6 is officially released? can I start packaging with the current source in branch? 2008-01-28 01:02 I think later today 2008-01-28 01:03 hopefully 2008-01-28 01:03 errr that's quite ambiguous to me :-) 2008-01-28 01:03 in hours? 2008-01-28 01:03 (I mean, it's 10am here, what time is it wherever you are?) 2008-01-28 01:04 ".6 branched, will release Monday or so. Trunk open for 0.7" 2008-01-28 08:19 -!- maximadelman(~maxim@smtp.violintech.net) has joined #zumastor 2008-01-28 09:45 -!- cbsmith(~xman@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-28 10:13 -!- maximadelman(~maxim@smtp.violintech.net) has left #zumastor 2008-01-28 11:25 -!- shapor(~shapor@yzf.shapor.com) has joined #zumastor 2008-01-28 11:32 ACTION is back 2008-01-28 11:36 ACTION is back 2008-01-28 12:07 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-28 14:10 can I trust what I read in the howto? it still talks about zumastor 0.4 and many changes went into 0.5 and 0.6 2008-01-28 14:37 pgquiles: None of the changes I can think of were user-level visible. 2008-01-28 14:37 pgquiles: We've mostly been focused on stability. 2008-01-28 14:38 cbsmith: thanks 2008-01-28 14:38 pgquiles: For focusing on stability or answering your question? ;-) 2008-01-28 14:38 :-P 2008-01-28 15:19 -!- kfortin_(~chatzilla@smtp.violintech.net) has joined #zumastor 2008-01-28 21:47 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-28 23:51 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Tue Jan 29 00:00:01 PST 2008 2008-01-29 10:24 -!- jiayingz(~jiayingz@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-29 10:35 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-29 10:55 -!- dkegel(~chatzilla@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-29 10:55 Daniel? 2008-01-29 12:07 hi dkegel 2008-01-29 12:08 note: the tab doesn't light up unless you type my exact nick 2008-01-29 12:42 flips, dkegel: after reading the howto and the man page, there are a few things which are not clear to me yet 2008-01-29 12:43 should I ask here or send an e-mail to the mailing list so that the answers are preserved for other people in the future? 2008-01-29 12:46 ok. following the 'never ask to ask' rule, here they come :-) 2008-01-29 12:48 1) zumastor define volume: is it mandatory that the snapshots are stored in an independent volume. I think someone asked this same question a few versions ago and the answer was 'yes, in the future it will be possible to have snapshots in the same volume'. Has that future arrived? 2008-01-29 12:49 pgquiles, you are welcome to submit patches to the howto 2008-01-29 12:49 2) zumastor define source / define target: do I need to use both (one on the master, the other on the destination) or can I just use either one of them and be done? 2008-01-29 12:50 flips: actually, I have questions not answers so no patches yet :-) 2008-01-29 12:50 and we try to keep it updated, but it is hard to give everything the attention it deserves 2008-01-29 12:50 sure 2008-01-29 12:50 send a mail is the right thing, indeed 2008-01-29 12:50 pgquiles: 1) yes 2) yes, you need both 2008-01-29 12:50 the short answer :) 2008-01-29 12:50 unfortunately, google is not indexing our group very well as I found yesterday 2008-01-29 12:50 we ought to be able to do something about that ;) 2008-01-29 12:51 3) zumastor snapshot: the howto sas "for manual snapshots, the usecount will drop to zero and the device will not be created after zumastor is restarted". That seems odd! if I create a snapshot manually, probably it's because I'm *really* interested in the data in that particular moment, so the least I expect is to be able to access that snapshot if I reboot 2008-01-29 12:52 pgquiles: 3) i agree :) 2008-01-29 12:52 flips: yeah, if google employees cannot achieve that, we might as weel turn the Internet off :-) 2008-01-29 12:52 you can use the usecount command to make it stick 2008-01-29 12:52 shapor: mmm no 'zumastor usecount' in my manpage :-? 2008-01-29 12:53 yeah, its not in zumastor yet, you could use ddsnap 2008-01-29 12:53 ah, ok 2008-01-29 12:53 but i agree that manual snapshots should be given a usecount 2008-01-29 12:54 otherwise whats the point 2008-01-29 12:54 pgquiles, re (1) the future, I think I will add it to the list of requirements for my lvm rewrite effort 2008-01-29 12:54 then there's the last question: is the second example in the manpage correct? I can't see how 'zumastor define master test -h 24 -d 7' means "keep the last 5 hours as snapshots'. To me, it means 'keep a snapshot for every hour for the last 7 days' (which would overflow the current 64 snapshots limit, btw) 2008-01-29 12:55 it just means keep 24 hourly snapshots (a days worth) and 7 daily snapshots (a weeks worth) 2008-01-29 12:55 by the way, the irc channel is logged on the web, but still the mailing list is a better place for Q&A 2008-01-29 12:55 yes, I was wrong 2008-01-29 12:56 flips: I agree 2008-01-29 12:56 I'll copy & paste this conversation to the m-l just for archiving purposes 2008-01-29 12:57 so, no "keep the last 5 hours as snapshots" for that second example, right? 2008-01-29 12:59 ah yeah that 5 hours thing is just plain wrong 2008-01-29 12:59 that would be -h 5 2008-01-29 12:59 yeah 2008-01-29 12:59 which would be a less confusing example than -h 24 2008-01-29 13:00 yup 2008-01-29 13:00 btw, what size is jianying's failing replicating volume? 2008-01-29 13:00 --hourly 5 would be even more clear 2008-01-29 13:01 jiayingz: ping 2008-01-29 13:02 I'm going to test zumastor tomorrow with 10x750GB disks 2008-01-29 13:03 if there's any testing you'd like me to perform, just tell me 2008-01-29 13:42 hi pgquiles 2008-01-29 13:43 we are copying 300G data 2008-01-29 13:43 300G? :-D 2008-01-29 13:43 I can test 20 times that :-) 2008-01-29 13:43 the bug we are looking at now is caused by the smalll snapshot store 2008-01-29 13:43 you mean the 64 snapshot limit? 2008-01-29 13:43 I am using 350G as origin and 50G as snapshot 2008-01-29 13:44 so you may not see the problem if u use a large snapshot volume 2008-01-29 13:44 we are working on fixing the bug now 2008-01-29 13:44 yes, I've read the issue tracker 2008-01-29 13:45 is there any advice on the size of the snapshot volume related to the size of the origin volume? 2008-01-29 13:45 hopefully we will have the fix in today or tommorrow 2008-01-29 13:45 cool :-) 2008-01-29 13:45 usually we set the size of the snapshot store to be larger than the origin volume 2008-01-29 13:46 just to make sure the snapshot store won't be full even if all of the origin data is overwritten 2008-01-29 13:46 but it is hard to do this with a small disk 2008-01-29 13:46 so we just make a large origin volume and a small snapshot store 2008-01-29 13:47 the short answer to your question is that depends on your workload 2008-01-29 13:47 what happens the snapshot volume is full and there are still snapshots to go to reach the max number of snapshots I've setup with 'zumastor define master'? Does zumastor remove the oldest snapshot, then try to perform the new snapshot, or is the new snapshot skipped? 2008-01-29 13:48 the snapshot store should be large enough so that old snapshots won't be automatically deleted under churns 2008-01-29 13:48 yes. zumastor will remove the old snapshots and take the new snapshot 2008-01-29 13:49 it will be nice to try the large-volume copy test after you set up zumastor 2008-01-29 13:49 to make sure zumastor works properly for that large volume 2008-01-29 13:49 is the step of removing the old snapshots to take the new snapshots logged? I mean, it'd be nice to be able to analyze the logs and know I need to resize the snapshots volume 2008-01-29 13:50 jiayingz: sure 2008-01-29 13:50 it is logged by ddsnap. under /var/log/zumastor/$volume/server.log 2008-01-29 13:51 jiayingz: what size do you want me to allocate for the origin and for the snapshots? is 3TB for the origin and 4.5GB for the snapshots ok? 2008-01-29 13:51 err 4.5TB 2008-01-29 13:51 that should be enough. 2008-01-29 13:52 good 2008-01-29 13:53 we tried 2T volume setup on qemu, but haven't tried that large volume test on real machine 2008-01-29 13:53 that will be great to know how it works with your setup. :) 2008-01-29 13:54 is there any script to generate data to fill the 3TB, then modify files randomly or should I write it myself? 2008-01-29 13:54 (I've seen something about zsoftware in your script) 2008-01-29 13:55 it can be any data that is large enough. zsoftware is the name of the mount point on a nfs client 2008-01-29 13:56 the variable to set what the data to be copied is $copy_path 2008-01-29 13:57 the test script is http://zumastor.googlecode.com/svn/trunk/test/large_volume_copy_test.sh 2008-01-29 13:57 A "createdata" verb would be great to add to large_volume_copy_test.sh 2008-01-29 13:58 one additional question: will zumastor take a touch-ed file as a different file? (ie. does it consider timestamps or sums for the file?) 2008-01-29 13:58 "for the files", I meant 2008-01-29 14:00 zumastor doesn't use checksums. It will notice that the metadata block has changed and propagate just that. 2008-01-29 14:00 sorry, it will notice that the metadata block has been written to, even if it hasn't changed. 2008-01-29 14:01 Does that make sense? 2008-01-29 14:02 pretty much :-) 2008-01-29 14:02 it means 'touch file' will take a add that file to the next snapshot 2008-01-29 14:02 which is what I wanted to know :-) 2008-01-29 14:03 BTW I'm very happy to hear you're willing and able to do a really large test. Will N. is going to do one, too, but he doesn't have the disks yet. 2008-01-29 14:04 pgquiles, just touching a file will indeed propagate it, but it isn't much of a stress test... it's just a single block change. 2008-01-29 14:05 dkegel: I have had the disks and servers for almost two months but I had other priorities :-( 2008-01-29 14:05 now I have time and hardware 2008-01-29 14:05 :-) 2008-01-29 14:09 " 2008-01-29 14:09 While ZFS and NILFS are filesystems, Zumastor is combined kernelspace/userspace approach which works on Linux on unmodified filesystems (you can use ext3, XFS, etc). 2008-01-29 14:09 2008-01-29 14:09 I would immediately go for ZFS (quite a lot of reasons actually, it just beats all the other filesystems available :/ ) but well, that Zumastor sounds like an interesting idea too. I had never heard of that so thank you for the link ;) Might start playing around with it, just out of curiosity, maybe even create a GUI ;) 2008-01-29 14:09 " 2008-01-29 14:09 in OSNews: http://osnews.com/comments/19232 2008-01-29 14:11 that will be awesome, pgquiles 2008-01-29 14:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-29 14:13 indeed 2008-01-29 14:13 I only hope she develops the GUI in English, not in Finnish (her website is all in Finnish :-) 2008-01-29 17:12 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-29 18:52 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-29 20:13 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-29 20:24 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-29 20:41 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor 2008-01-29 22:11 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-29 22:12 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has left #zumastor 2008-01-30 12:16 -!- zumalog(~zumalog@yzf.shapor.com) has joined #zumastor 2008-01-30 12:57 dank: its fixed now 2008-01-30 13:32 Thanks... 2008-01-30 13:34 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-30 14:32 -!- pgquiles_(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor 2008-01-30 14:33 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-30 14:57 -!- charlesnw(~charles@dctm.siderean.com) has joined #zumastor 2008-01-30 18:08 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-30 19:07 -!- natalie(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-30 19:11 -!- nataliep(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-30 20:58 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-30 21:08 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-30 21:43 -!- nataliep(~nataliep@cpe-76-94-49-21.socal.res.rr.com) has joined #zumastor 2008-01-30 21:50 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-30 23:21 -!- pgquiles(~pgquiles@62.43.226.52.static.user.ono.com) has joined #zumastor irc.oftc.net #zumastor log beginning Thu Jan 31 00:00:01 PST 2008 2008-01-31 07:19 -!- charlesnw(~charles@ses.siderean.com) has joined #zumastor 2008-01-31 07:37 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-31 07:40 -!- tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-31 08:49 -!- mitchg(~mitchg@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-31 10:14 -!- charlesn1(~charles@ses.siderean.com) has joined #zumastor 2008-01-31 10:32 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-31 10:38 -!- daniel__(~phlipz@phunq.net) has joined #zumastor 2008-01-31 10:42 -!- pgquiles(~pgquiles@81.202.65.108.dyn.user.ono.com) has joined #zumastor 2008-01-31 11:34 pgquiles, there? 2008-01-31 11:36 flipz: hi 2008-01-31 11:37 pgquiles, I am putting together a proposal to handle your exact replication use case 2008-01-31 11:37 I hope 2008-01-31 11:37 that'd be great! 2008-01-31 11:37 where you want to get the same snapshot out to every downstream server 2008-01-31 11:37 I see named snapshots as the way to do it 2008-01-31 11:37 I've read this morning about your meeting but I could not pay as much attention as I wanted because I had a fire to extinguish 2008-01-31 11:38 on master you give a command like zumastor snapshot --name golden-1 2008-01-31 11:38 ah 2008-01-31 11:38 I don't see the need for names 2008-01-31 11:38 it is a convenient way to say that that particular snapshot is special 2008-01-31 11:39 and also lets you see easily when that snapshot has made it to downstream clients 2008-01-31 11:39 it is easy for cron to autogenerate the name 2008-01-31 11:39 mmm yes 2008-01-31 11:39 for example: snap-$(date) 2008-01-31 11:40 but it's not the use case I'd like to have (not that you have to work for me :-) 2008-01-31 11:40 so you put zumastor snapshot myvolume --name snap-$(date) in your cron job 2008-01-31 11:40 I want to have the same snapshots in the origin and in the replicas, plan and simple 2008-01-31 11:40 could you describe your use case again please 2008-01-31 11:40 thanks 2008-01-31 11:40 the command I showed above will do it 2008-01-31 11:41 if I put that in cron, I no longer need zumastor define master myvol -d 30, etc, right? 2008-01-31 11:41 right 2008-01-31 11:42 but you can also put a similar thing in cron.hourly if you want 2008-01-31 11:42 it's up to you 2008-01-31 11:42 wait, my command syntax was not quite right 2008-01-31 11:43 but with manual snapshots I have the problem the device will not be created after zumastor is restarted, right? 2008-01-31 11:43 so those snapshots won't be available after a reboot, for instance 2008-01-31 11:44 we use the "custom" kind of snapshot for that 2008-01-31 11:44 I'm just reading the man page to get the syntax right ;) 2008-01-31 11:44 :-) 2008-01-31 11:44 snapshot myvolume custom mysnaps --name snap-$(date) 2008-01-31 11:45 and: define master --custom mysnaps:8 2008-01-31 11:45 which puts a limit on custom snapshots that can be retained 2008-01-31 11:45 you can also have no limit 2008-01-31 11:45 and just rely on autodelete 2008-01-31 11:46 fine, that gives me named snapshots 2008-01-31 11:46 how do I get them replicated downstream? 2008-01-31 11:46 it just happens, that is part of my proposal 2008-01-31 11:47 zumastor replicate -s mysnapname ? 2008-01-31 11:47 err 2008-01-31 11:47 zumastor replicate volume host -s mysnapname 2008-01-31 11:47 zumastor just does that as soon as you define a target 2008-01-31 11:47 I meant 2008-01-31 11:47 by default 2008-01-31 11:47 ah ok 2008-01-31 11:47 happy so far? 2008-01-31 11:48 very happy :-) 2008-01-31 11:48 ok, I will write the proposal 2008-01-31 11:48 there's still one thing to think about: what happens if you try to replicate a snapshot while the previous replication is still going on 2008-01-31 11:49 queued replications? cancelled replications? 2008-01-31 11:57 -!- charlesnw(~charles@dctm.siderean.com) has joined #zumastor 2008-01-31 11:58 currently we just continue sending and ignore the request 2008-01-31 11:58 theres no queueing 2008-01-31 12:00 shapor: do you mean the second replication is just discarded? (but it's logged, I assume) 2008-01-31 12:01 yes 2008-01-31 12:01 ACTION just filed two usability-related bugs 2008-01-31 12:01 ok 2008-01-31 12:05 btw, will 'zumastor replicate' commit a change in downstream to upstream? i. e. if A replicates to B, and I modify a file in B, is there any way to have that change in A too? 2008-01-31 12:06 pgquiles: good question, many people would like that feature 2008-01-31 12:06 we've been thinking about it (well flipz mostly) 2008-01-31 12:06 but the short answer is no 2008-01-31 12:08 I wonder if this would work: 2008-01-31 12:08 in B, 'define source myvol A' 2008-01-31 12:08 in A, 'define target myvol B' 2008-01-31 12:08 in A, 'define source myvol B' 2008-01-31 12:09 in B, 'define target myvol A' 2008-01-31 12:09 mmm probably a bit of a mess, I'd need to test that tomorrow :-) 2008-01-31 12:09 that wont work 2008-01-31 12:10 the script wont let you 2008-01-31 12:10 and if we let it, it would corrupt the filesystem 2008-01-31 12:11 :-) 2008-01-31 12:57 What does one have to poke to get a replication target to mount a volume 2008-01-31 13:28 I guess I am just waiting for initial replication. 2008-01-31 14:14 willn, we don't automatically mount snapshots on downstream 2008-01-31 14:15 except for the latest 2008-01-31 14:15 hmm 2008-01-31 14:15 double hmm 2008-01-31 14:16 pgquiles, propagating changes on downstream back to upstream is Hard[tm] 2008-01-31 14:16 but there are ideas how to do it 2008-01-31 14:16 flipz: :-) 2008-01-31 14:17 when you reach zumastor 1.0 you can announce it as "the zumastor team is very happy to release zumastor 1.0 *in spite of pgquiles asking once and again for crazy hard features*" :-P 2008-01-31 14:17 shapor, we should probably remember that somebody asked for a new replication while one was already going, and only forget the request completely if we have already remembered one 2008-01-31 14:17 pgquiles, you are not the first to ask for that 2008-01-31 14:18 the second in fact 2008-01-31 14:18 shapor, ping 2008-01-31 14:21 pong 2008-01-31 14:21 flipz 2008-01-31 14:28 flipz: I was hoping for *something* to be mounted. But I think i'm still waiting for the initial snapshot transfer (500G @ ~100Mb/s) 2008-01-31 14:29 something is supposed to be mounted 2008-01-31 14:29 Well. hmm 2008-01-31 14:29 shapor has a replication monitoring script 2008-01-31 14:29 showing progress 2008-01-31 14:29 zuma6 is supposed to be replicating to zuma5 if you want to poke at them 2008-01-31 14:29 nothing will be mounted until the initial replication cycle completes, perhaps we ought to rethink that 2008-01-31 14:30 I think 2008-01-31 14:30 haven't looked at that part of zumastor script for a long time 2008-01-31 14:30 ah right, that would make sense then. 2008-01-31 15:02 its in the repository under zumastor/scripts iirc 2008-01-31 16:38 -!- Tim_vimm(~Tim@cpe-76-90-128-140.socal.res.rr.com) has joined #zumastor 2008-01-31 17:19 -!- charlesnw(~charles@dctm.siderean.com) has left #zumastor 2008-01-31 18:51 -!- nataliep(~nataliep@207.47.98.129.static.nextweb.net) has joined #zumastor 2008-01-31 19:27 -!- charlesnw(~charles@cpe-75-84-92-236.socal.res.rr.com) has joined #zumastor 2008-01-31 19:46 -!- charlesnw(~charles@cpe-75-84-92-236.socal.res.rr.com) has joined #zumastor 2008-01-31 19:52 -!- nataliep(~nataliep@cpe-76-94-49-21.socal.res.rr.com) has joined #zumastor 2008-01-31 21:17 -!- charlesn1(~charles@cpe-75-84-92-236.socal.res.rr.com) has joined #zumastor 2008-01-31 21:17 -!- charlesn1(~charles@cpe-75-84-92-236.socal.res.rr.com) has left #zumastor 2008-01-31 22:13 -!- charlesnw(~charles@cpe-75-84-92-236.socal.res.rr.com) has joined #zumastor 2008-01-31 22:21 http://phunq.net/fit.jpg 2008-01-31 22:47 -!- cbsmith(~xman@adsl-99-165-23-73.dsl.lsan03.sbcglobal.net) has joined #zumastor