Camelia, the Perl 6 bug

IRC log for #gluster-dev, 2013-02-07

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:52 overclk joined #gluster-dev
02:57 __Bryan__ joined #gluster-dev
03:18 bharata joined #gluster-dev
03:26 bulde joined #gluster-dev
03:53 sripathi joined #gluster-dev
04:05 sgowda joined #gluster-dev
04:12 pai joined #gluster-dev
04:27 bala1 joined #gluster-dev
04:33 sripathi1 joined #gluster-dev
04:36 sripathi joined #gluster-dev
04:38 deepakcs joined #gluster-dev
05:05 vpshastry joined #gluster-dev
05:26 mohankumar joined #gluster-dev
06:05 raghu joined #gluster-dev
06:07 sripathi1 joined #gluster-dev
06:33 sripathi joined #gluster-dev
06:39 bharata_ joined #gluster-dev
06:56 kanagaraj joined #gluster-dev
07:00 sripathi joined #gluster-dev
07:05 sripathi joined #gluster-dev
07:58 puebele joined #gluster-dev
08:34 deepakcs joined #gluster-dev
09:05 sgowda joined #gluster-dev
09:30 sripathi1 joined #gluster-dev
09:46 blues-man joined #gluster-dev
09:50 sripathi joined #gluster-dev
09:52 _benoit_ joined #gluster-dev
10:52 bala1 joined #gluster-dev
11:29 hagarth joined #gluster-dev
11:53 bala1 joined #gluster-dev
12:01 kkeithley1 joined #gluster-dev
12:08 bluesman_ joined #gluster-dev
12:52 edward1 joined #gluster-dev
13:23 __Bryan__ left #gluster-dev
14:25 deepakcs joined #gluster-dev
14:30 mohankumar joined #gluster-dev
14:34 bharata joined #gluster-dev
15:21 vpshastry joined #gluster-dev
15:28 wushudoin joined #gluster-dev
15:28 vpshastry1 joined #gluster-dev
16:01 16WAAD817 joined #gluster-dev
16:25 hagarth joined #gluster-dev
16:30 portante joined #gluster-dev
16:33 vpshastry1 left #gluster-dev
16:45 xavih I've published an initial version of the disperse translator on github (https://github.com/datalab-bcn)
16:46 xavih it's very preliminar but it seems to be working on normal cases
16:46 xavih it requires another translator also published on github
16:47 xavih the README explains the most important things
16:48 joshcarter joined #gluster-dev
16:50 joshcarter I'm looking for some help with protocol/server and storage/posix behavior on lookup. I have an xlator somewhat similar to stripe, in that it has children with the same file and matching GFID's….
16:51 joshcarter One child is looked up first, the other might or might not be looked up. On Gluster 3.3, the server crashes if I do a lookup on a loc which has the GFID filled in.
16:52 joshcarter So, I'm trying to figure out: do I need to specially prepare (or save off) a loc which is identical to the one I got at lookup time, or can I figure out how the server's mis-behaving in this situation.
17:12 ndevos xavih: maybe you can add that link to http://www.gluster.org/community/docum​entation/index.php/Planning34/Disperse ?
17:19 xavih ndevos: I've already added it, at the bottom though...
17:21 johnmark xavih: sweet!
17:22 johnmark xavih: on another note, I owe you something. coming right up...
17:22 xavih johnmark: no problem :)
17:25 semiosis johnmark: pdurbin was looking for you in #gluster earlier
17:26 semiosis also, https://launchpad.net/~semiosis​/+archive/ubuntu-glusterfs-3.4
17:28 johnmark semiosis: \m/
17:28 johnmark semiosis: oh? ok
17:30 johnmark kkeithley1: ping
17:31 johnmark xavih: would you like to send a note to gluster-devel announcing it? only has to be brief. I'm happy to review it if that would be best
17:33 johnmark kkeithley: ping
17:33 * johnmark isn't sure which of kkeithley's sock puppets are around
17:41 hagarth joined #gluster-dev
18:43 kkeithley1 johnmark: poing
18:44 kkeithley1 sorry, gnome3 was hiding my irc app
18:45 glusterbot joined #gluster-dev
18:49 johnmark ah, no worries
18:49 johnmark kkeithley1: just wanted to nag you for alpha build :)
18:49 kkeithley1 ah
18:49 johnmark kkeithley1: oh sure, *everything* gets blamed on gnome 3 ;)
19:01 a2 joshcarter, what kind of crash did you see? can you provide more details of what you are trying to do in the lookup?
19:07 kkeithley1 a2, when a verification gets fired off on build.g.o, where does the creation of the glusterfs-3git.tar.gz happen? (Not the scripts in /opt/qa or /opt/engg/bin obviously)
19:07 joshcarter a2: sorry, I should have saved some stack traces. It had gone through posix_lookup ok, and was unwinding and cleaning up inodes. In there, it appeared to be double-unref'ing an inode.
19:08 joshcarter a2: I wound up skirting the problem by saving off a copy of the loc_t that I get at lookup time, and using that for my deferred lookup on the secondary child.
19:10 joshcarter (the double-unref is an assumption. It was pruning the inode table, and table->lru_limit had gone into 0xffff… range. Anyway, not a problem for me anymore, but if you're interested I can repro the problem with an older version of my code.
19:12 kkeithley1 johnmark: I'll work on it tomorrow when I'm snowed in. Wo
19:13 kkeithley1 Your first Nor'easter.
19:15 johnmark woohoo :)
19:18 johnmark kkeithley1: I'm pushing a blog post this afternoon. Note that the only repo-able builds we have right now are for Ubuntu
19:18 semiosis kkeithley1: fyi, i'll be afk tmrw, going to sunshinephp conf, so unable to assist if you need help
19:19 johnmark kkeithley1: which doesn't bother me, but for future alphas/betas it would be super nice if we could handle this automagically
19:19 johnmark can we get you, jdarcy, hagarth and a2 in a room to figure this out?
19:19 johnmark or maybe we'll just wait for dev summit
19:51 nixpanic johnmark: have you seen the oVirt/Gluster persentation from http://simon3z.blogspot.com/201​3/02/ovirt-at-fosdem-2013.html ?
19:53 kkeithley1 semiosis: you mean (can't) help with building .debs for debian. I actually wasn't even thinking of building those at this point in time. And I thought we had a volunteer who was going to take over builds for debian anyway.
19:54 semiosis oh oops thought thats what you & johnmark were talking about
19:54 semiosis whos the volunteer?
19:58 kkeithley1 Since johnmark asked yesterday whether the alpha rpms were built with the Fedora .spec I sorta presumed that's what he was after. I don't remember — off the top of my head — who the volunteer was.
19:59 johnmark kkeithley1: yes, that's what I was after
19:59 johnmark kkeithley1: er, we had a .deb volunteer?
20:01 a2 sorry got pulled into a call
20:02 * a2 reading messages
20:02 a2 kkeithley, are you talking about the RPM verification?
20:03 a2 joshcarter, is your code shareable? it might be extra unrefs
20:05 kkeithley1 a2: I'm talking about the gerrit/jenkins build verification, including the rpm.t test that's run. My new test needs both the glusterfs-3git.tar.gz and the ufo tarball to build the RPMs necessary to pass the test
20:05 joshcarter a2: sorry, no, company-internal for now. (it's a prototype, not something that's shipped.)
20:08 kkeithley1 During the test run, where things appear to be unpacked in /var/lib/jenkins/jobs/regression/workspace/ the glusterfs-3git.tar.gz gets made.
20:09 a2 kkeithley, glusterfs-3.git is never made except in rpm.t
20:09 a2 glusterfs-3git.tar.gz I meant
20:09 kkeithley1 Ah,
20:10 kkeithley1 Right, the `make dist` in rpm.t.
20:13 kkeithley1 So /var/lib/jenkins/.../workspace is a red herring. I'm trying to find out more about what's failing, available logs aren't giving me much. (I saw that when I triggered a verification run that /var/lib/jenkins/.../workspace has my changes in it.)
20:13 kkeithley1 So I'm looking for some tips on where to look
20:15 kkeithley1 for more info
20:38 badone joined #gluster-dev
20:42 a2 all your clues are necessarily from within rpm.t as your starting point
20:43 a2 you could, pause jenkins from running new jobs, and manually execute rpm.t from the workspace directory as root
20:43 a2 (and set some env variables like $PATH to /build/install/sbin)
21:03 hagarth joined #gluster-dev
22:01 kkeithley1 uh yeah, looks like I have to do this the hard way

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary