Camelia, the Perl 6 bug

IRC log for #gluster, 2013-08-24

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

All times shown according to UTC.

Time Nick Message
00:02 thanosme joined #gluster
00:20 JerryM left #gluster
01:32 vpshastry joined #gluster
01:46 jporterfield joined #gluster
01:52 jporterfield joined #gluster
01:58 jporterfield joined #gluster
02:04 vpshastry joined #gluster
02:10 jporterfield joined #gluster
02:16 Humble joined #gluster
02:29 bala joined #gluster
02:34 ultrabizweb joined #gluster
02:36 jporterfield joined #gluster
02:37 nueces joined #gluster
03:17 bulde joined #gluster
03:45 bulde joined #gluster
03:56 jporterfield joined #gluster
04:10 smellis joined #gluster
04:10 awheeler joined #gluster
04:12 smellis i've got a problem with a 2 brick replicated volume on 3.3.1, I'm seeing the same file with a bunch of different date in the info split-brain output
04:13 smellis I followed joe julian's guide to fix sb on 3.3, but it's still there
04:21 hagarth joined #gluster
04:28 jporterfield joined #gluster
04:41 bulde joined #gluster
04:43 jporterfield joined #gluster
05:03 jporterfield joined #gluster
05:18 vpshastry joined #gluster
05:18 vpshastry left #gluster
05:19 jporterfield joined #gluster
05:28 jporterfield joined #gluster
05:28 bala joined #gluster
05:40 jporterfield joined #gluster
05:52 jporterfield joined #gluster
05:58 jporterfield joined #gluster
06:18 jporterfield joined #gluster
06:24 dusmant joined #gluster
06:26 jporterfield joined #gluster
06:32 jporterfield joined #gluster
06:45 jporterfield joined #gluster
06:49 _pol joined #gluster
07:10 jporterfield joined #gluster
08:00 jporterfield joined #gluster
09:57 neofob joined #gluster
11:10 jporterfield joined #gluster
11:18 jporterfield joined #gluster
11:31 ababu joined #gluster
11:32 jporterfield joined #gluster
11:38 jporterfield joined #gluster
11:40 bulde joined #gluster
12:05 Rydekull joined #gluster
12:18 Rydekull So, anyone got a suggestion? :-) http://pastebin.com/bPeX8AZk
12:18 glusterbot Please use http://fpaste.org or http://paste.ubuntu.com/ . pb has too many ads. Say @paste in channel for info about paste utils.
12:29 hagarth Rydekull: the underlying file system doesn't seem to support extended attributes?
12:29 jporterfield joined #gluster
12:35 jporterfield joined #gluster
12:35 Rydekull Well, i agree thus far
12:35 Rydekull trying to figure out why though
12:43 jporterfield joined #gluster
12:47 Rydekull It just seems illogical
12:48 Rydekull It must be something else then extended attributes
12:48 Rydekull Hrm
12:56 bala joined #gluster
13:03 bala1 joined #gluster
13:14 Rydekull right, so, zfs-fuse = no good for gluster, zfs from zfsonlinux = good for gluster
13:15 jporterfield joined #gluster
13:21 jporterfield joined #gluster
13:30 jporterfield joined #gluster
13:54 lalatenduM joined #gluster
14:01 mohankumar joined #gluster
14:04 jporterfield joined #gluster
14:13 robo joined #gluster
14:29 jporterfield joined #gluster
14:29 hagarth joined #gluster
14:33 smellis what does it mean when there are multiple split brain entries for the same file, but only on one host?
14:35 smellis http://pastebin.com/aHtNBYhX
14:35 glusterbot Please use http://fpaste.org or http://paste.ubuntu.com/ . pb has too many ads. Say @paste in channel for info about paste utils.
14:36 morsik ads? ;o
14:36 smellis ah ok
14:36 smellis I'll paste it there
14:36 morsik smellis: it's bot -.-
14:37 smellis http://ur1.ca/f73op
14:37 glusterbot Title: #34426 Fedora Project Pastebin (at ur1.ca)
14:37 smellis sorry for the violation of etiquette
14:37 smellis lol
14:41 smellis went back and looked at pastebin, I realized I just naturally filter our ads in my head
14:41 smellis I didn't even notice them the first time
14:44 smellis made a backup of that file and it's still showing up
14:44 smellis in the sb info output
14:45 ndevos smellis: that list is like a journal, or log of the split-brains that were detected, you probably want to check the /var/log/glusterfs/glustershd.log (shd: self heal daemon) file
14:49 smellis roger, will do
14:49 smellis I made a backup of the good copy of the file, then deleted from the gluster volume, but it still shows
14:50 smellis how do I remove it completely?
14:50 smellis docs are pretty sparse in this area
14:53 glusterbot New news from newglusterbugs: [Bug 986429] Backupvolfile server option should work internal to GlusterFS framework <http://goo.gl/xSA6n>
14:53 smellis I should say, it still shows in the info sb output
15:00 jporterfield joined #gluster
15:05 smellis crap, I just noticed that 4 other files are listed as heal failed
15:05 ndevos ~split brain | smellis
15:05 glusterbot smellis: I do not know about 'split brain', but I do know about these similar topics: 'split-brain', 'splitbrain'
15:06 ndevos ~split-brain | smellis
15:06 glusterbot smellis: To heal split-brain in 3.3+, see http://goo.gl/FPFUX .
15:06 ndevos @splitbrain
15:06 glusterbot ndevos: To heal split-brain in 3.3+, see http://goo.gl/FPFUX .
15:06 ndevos oh, it's a dup!
15:07 ndevos good luck smellis, I'm out for the day
15:07 smellis yup, did that
15:07 smellis thanks, have a good one
15:19 jporterfield joined #gluster
15:27 glusterbot New news from resolvedglusterbugs: [Bug 764966] gerrit integration fixes <http://goo.gl/AZDsh>
15:27 jporterfield joined #gluster
15:29 JoeJulian smellis: Those entries are from June...
15:31 smellis right, but what I don't get is why they are still there, and specifically only on hyper2
15:31 smellis if it's split brain, shouldn't both nodes have an entry?
15:32 smellis I guess I just don't understand the output of that command
15:33 smellis when I see entries in the sb or heal failed output, how do I know if something is actually wrong or not (short of taking md5sums on the backing store)?
15:34 JoeJulian It's there because back in June, it detected a split brain from hyper2's self-heal daemon and logged it.
15:35 smellis ok, so that's just parsing the logs?
15:35 davinder joined #gluster
15:35 davinder Hi
15:35 glusterbot davinder: Despite the fact that friendly greetings are nice, please ask your question. Carefully identify your problem in such a way that when a volunteer has a few minutes, they can offer you a potential solution. These are volunteers, so be patient. Answers may come in a few minutes, or may take hours. If you're still in the channel, someone will eventually offer an answer.
15:35 JoeJulian ,,(meh) I think it should be able to be cleared as well. The only way I know of clearing it is to restart glusterd(s).
15:35 glusterbot I'm not happy about it either
15:35 smellis ah, yeah I read about that in the mailing list, I'm a dummy and didn't put that together myself
15:36 smellis so, imagine I have two hypervisors, running VMs, but also hosting bricks for a replicate volume
15:36 smellis how do I know the storage is synced up and valid on both machines?
15:37 JoeJulian I'll try, but it's the weekend and I haven't had coffee yet...
15:37 smellis haha
15:37 smellis i understand
15:38 smellis I appreciate your blog post on fixing sb by the way :)
15:38 davinder I have one confusion ...if I mount gluster on client then if one gluster server down how client will auto mount on other replica of volume ??
15:38 JoeJulian Monitor your client logs for errors. Check "gluster volume heal $vol info". If you're REALLY paranoid, compare hashes and check the ,,(extended attributes).
15:38 glusterbot (#1) To read the extended attributes on the server: getfattr -m .  -d -e hex {filename}, or (#2) For more information on how GlusterFS uses extended attributes, see this article: http://goo.gl/Bf9Er
15:38 smellis davinder: the clients connect to all the bricks directly (if your using the native client)
15:39 JoeJulian ~mount server | davinder
15:39 glusterbot davinder: The server specified is only used to retrieve the client volume definition. Once connected, the client connects to all the servers in the volume. See also @rrnds
15:39 davinder its a SUSE 11 so using mount command
15:39 smellis davinder: are you using the fuse client, i.e. -t glusterfs on your mount?
15:40 davinder not fuse client
15:40 davinder NFS mount
15:40 smellis davinder: aha, you'll have to make the address you mount to highly available in that case
15:40 davinder i didnt get SuSe fuse client for 11 & 10
15:41 JoeJulian nfs doesn't do that. You'll have to use another tool to create a virtual IP address that will move around as needed.
15:41 davinder i have to create hearbeat or HA setup for that ?
15:41 smellis davinder: in other words, you need to use carp or vrrp or rhcs
15:41 JoeJulian Either that, or use the fuse client.
15:41 smellis second that, it's kinda what make gluster really awesome
15:41 _pol joined #gluster
15:42 smellis davinder: checkout keepalived for making a vip between your brick nodes for nfs mount
15:42 smellis davinder: but I think nfs clients don't like to be disconnected and tend to hang
15:43 JoeJulian Hehe, was just saying the same thing: nfs mounts are TCP. If you fail-over you're going to get disconnected.
15:43 davinder It will not disconnect but if server dowm
15:45 davinder i hope keepalive or hearbeat will be good option for HA of NFS
15:46 JoeJulian Not in my opinion.
15:47 smellis yeah, there is really no good option for nfs
15:47 smellis I've had boxes that I've had to power off manually to get the nfs client to stop blocking
15:48 davinder ohhh
15:48 JoeJulian davinder: Why don't you want to use the fuse client?
15:49 davinder then i have to try to get for SuSe
15:49 davinder JoeJulian
15:50 hagarth joined #gluster
15:51 JoeJulian davinder: suse uses rpms. Why not just use the ,,(yum repo)
15:51 glusterbot davinder: The official community glusterfs packages for RHEL (including CentOS, SL, etc), Fedora 17 and earlier, and Fedora 18 arm/armhfp are available at http://goo.gl/s077x. The official community glusterfs packages for Fedora 18 and later are in the Fedora yum updates repository.
15:51 davinder any one  knows the offical path to download fuse client for SuSe 10 & 11
15:51 JoeJulian I'm pretty sure you can use the EL6 packages
15:52 davinder thanks JoeJulian
15:52 JoeJulian Oh, and your client version will need to match your server version.
15:53 JoeJulian (not entirely true as 3.3 and 3.4 seem to be compatible)
15:53 davinder i have 3.4
16:09 smellis so when there is sb detected on a file, what does gluster do until it's fixed?
16:10 smellis on other words, what data do clients get and which bricks are the data written to?
16:11 JoeJulian If a file is sb, the clients will get an error (EACCESS iirc) when trying to do anything with it.
16:11 smellis ah ok
16:44 bulde joined #gluster
16:44 _pol joined #gluster
17:13 jporterfield joined #gluster
17:29 jporterfield joined #gluster
18:19 toad joined #gluster
18:31 hagarth joined #gluster
20:21 awheeler joined #gluster
20:30 ricky-ticky joined #gluster
20:45 jporterfield joined #gluster
20:52 awheeler joined #gluster
21:14 awheeler joined #gluster
21:30 jporterfield joined #gluster
21:42 jporterfield joined #gluster
21:49 awheeler joined #gluster
21:55 jporterfield joined #gluster
22:16 jporterfield joined #gluster
22:22 jporterfield joined #gluster
22:41 Guest53741 joined #gluster
22:47 jporterfield joined #gluster
22:53 jporterfield joined #gluster
23:00 badone joined #gluster
23:18 jporterfield joined #gluster
23:31 _pol joined #gluster
23:47 jporterfield joined #gluster
23:56 jporterfield joined #gluster

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