Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster, 2013-11-09

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

All times shown according to UTC.

Time Nick Message
00:00 ira joined #gluster
00:06 elyograg nage: performance and possible or real file locking problems.  Lucene is the underlying technology behind both Solr and Elasticsearch.  It does not do well with network filesystems.  The problems with NFS and Lucene are legendary, both locking and performance.
00:09 elyograg it just works better with a real filesystem available, the faster the disk the better.  In truth people need to have enough OS disk cache to cache the entire index in RAM, so the disk is hardly ever hit.  I don't think there's any local caching of contents on the remote gluster filesystem, which blows THAT right out of the water.
00:10 vpshastry joined #gluster
00:10 elyograg i could be wrong about the caching thing, of course.
00:10 JoeJulian There would be if the FD isn't closed.
00:12 elyograg I don't know how Lucene does that.  Changing the index results in a new searcher object being created, so if that opens a new FD rather than reusing the existing one, sounds like that would kill it.  With local filesystems, the disk caching is block-level and persistent across multiple processes, threads, file opens/closes, etc.
00:16 badone joined #gluster
00:18 nage thanks for the explaination.  I'm working with a proprietary alternative to lucene, so not all the same constraints apply, but I'll have to go look into some of the lucene legends to find out what I should watch out for.
00:20 frostyfrog joined #gluster
00:48 mattappe_ joined #gluster
00:52 B21956 joined #gluster
00:54 calum_ joined #gluster
00:57 chriswonders Update on OS X: the source does not seem to compile on the Mac
00:58 chriswonders I found an archive from the gluster-users mailing list that seems to imply that it simply will not work on OS X
00:58 chriswonders http://comments.gmane.org/gmane.comp.file-systems.gluster.user/12826
00:58 glusterbot <http://goo.gl/Md81pO> (at comments.gmane.org)
01:00 chriswonders It's looking that it was once supported but is no longer.
01:10 bronaugh joined #gluster
01:10 chriswonders left #gluster
01:10 bronaugh hmm, so here's a fun one.
01:10 bronaugh [2013-11-09 01:03:25.948200] E [posix.c:4280:init] 0-climate-CMIP5-posix: mismatching volume-id (23573c6f-0788-456e-84c4-a511458939e3) received. already is a part of volume 0d24c8ad-50d9-468d-ba65-a5d37337aead
01:10 bronaugh this upon reboot.
01:10 bronaugh apparently, nothing changed.
01:12 dbruhn why did you reboot?
01:16 dbruhn You can manually edit your volume files on the server that is bad
01:23 mattappe_ joined #gluster
01:27 davidbierce joined #gluster
01:32 itisravi joined #gluster
01:32 bronaugh dbruhn: well; because we were patching a filesystem bug.
01:33 bronaugh zfs, specifically.
01:33 dbruhn ahh ok
01:33 dbruhn if you start checking the files in /var/lib/glusterd you will see the vol files in there
01:34 dbruhn you will want to figure out which file has the wrong volume id in it
01:34 bronaugh jesus, think I figured it out.
01:34 bronaugh don't think this is a glusterfs prob
01:34 dbruhn extended attributes get screwed up?
01:34 bronaugh kind of, kind of not. the underlying filesystems didn't get mounted.
01:34 dbruhn lol, that would do it
01:41 zaitcev joined #gluster
01:49 bronaugh yeah figured it out. yay? :/
01:49 bronaugh left #gluster
01:50 dbruhn That's a question? lol
02:20 davidbie_ joined #gluster
02:21 davidbierce joined #gluster
02:48 bala joined #gluster
02:50 itisravi joined #gluster
03:00 shubhendu joined #gluster
03:06 nasso joined #gluster
03:22 harish joined #gluster
03:39 elyograg any interactive help related to my failed rebalance email that just went to the user list would be appreciated.
03:39 elyograg i'll be afk for a bit here, but barring severe network problems, I'll see anything that's said.
03:42 itisravi_ joined #gluster
03:43 kanagaraj joined #gluster
03:56 RameshN joined #gluster
04:15 davidbierce joined #gluster
04:16 mohankumar joined #gluster
04:18 ndarshan joined #gluster
04:33 kshlm joined #gluster
04:44 aravindavk joined #gluster
05:03 paratai_ joined #gluster
05:04 bulde joined #gluster
05:06 bala joined #gluster
05:11 compbio joined #gluster
05:12 TDJACR joined #gluster
05:12 JoeJulian joined #gluster
05:12 dusmant joined #gluster
05:12 neofob_home joined #gluster
05:20 hflai_ joined #gluster
05:25 meghanam joined #gluster
05:25 meghanam_ joined #gluster
05:25 bala joined #gluster
05:41 ababu joined #gluster
05:42 bulde joined #gluster
08:12 purpleidea joined #gluster
08:12 purpleidea joined #gluster
08:40 bulde joined #gluster
09:02 ricky-ticky joined #gluster
09:20 davinder joined #gluster
09:21 diegows_ joined #gluster
09:42 ndarshan joined #gluster
09:51 calum_ joined #gluster
10:17 kbut joined #gluster
10:17 kbut Hi?
10:17 glusterbot kbut: 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.
10:18 kbut I have a problem with gluster client mount on 32 bits and acl's
10:19 kbut someone can help me?
10:39 kbut expose my problem: I have a 32 bits server ( I try with ubuntu and archlinux ) with a partition in xfs format. I create gluster volume named test1, later i start volume and i mount this with "mount -t glusterfs -o acl server-ip:/test1 /mnt" . Then i make directory myfolder on mnt. Change owner and group of myfolder to nobody:nogroup. Change permissions of myfolder to 770.  Finally set acl to myfolder to give all permissions to user1 user thr
10:39 kbut ough setfacl -m u:user1:rwx /mnt/myfolder
10:39 mohankumar joined #gluster
10:40 kbut But user1 can do anything over myfolder despite acl permission. Over 64 bits all work correctly, but over 32 not work.
10:43 kbut In 32 bits i can see acl's on myfolder, but seems that these not work. Please , can someone  reply something?
10:52 hateya joined #gluster
10:53 ndarshan joined #gluster
10:53 dusmant joined #gluster
10:54 RameshN joined #gluster
10:56 kanagaraj joined #gluster
10:58 shubhendu joined #gluster
11:02 ababu joined #gluster
11:09 ndevos kbut: I think you should file a bug about that, explain it as clearly as you have done here, and include command examples and exact package versions - it sounds quite reproducible to me
11:09 glusterbot http://goo.gl/UUuCq
11:11 kbut ndevos: i write bug now on bugtracker. But create new bug or reply in Bug 862908????
11:11 glusterbot Bug http://goo.gl/40xYSw unspecified, medium, ---, sgowda, ASSIGNED , glusterfs ACL permission denied problem
11:11 bala joined #gluster
11:17 ndevos kbut: right, can you reply to that bug? and have you tried with enble-ino32?
11:17 kbut yes, i try with enable-ino32, but result is same
11:18 mohankumar joined #gluster
11:22 kbut I just write reply. At any rate, can i do more test?
11:24 Alex Is running Gluster with unicode filenames a relatively uncommon thing btw? I don't see much on the Internet about it, and the bug I logged (https://bugzilla.redhat.com/show_bug.cgi?id=1024181) doesn't seem to happen to anyone else, so I'm stuck between 'misconfiguration' and actual defect..
11:24 glusterbot <http://goo.gl/wPFkNZ> (at bugzilla.redhat.com)
11:24 glusterbot Bug 1024181: unspecified, unspecified, ---, csaba, NEW , Unicode filenames cause directory listing interactions to hang/loop
11:24 davidbierce joined #gluster
11:29 ndevos kbut: I think that your reply is sufficient, you are now also CC'd on the bug, so you will get an email on an update or request for information
11:32 ndevos Alex: I dont think that is very common, I have not seen anyone do that before
11:36 mohankumar joined #gluster
11:36 Alex ndevos: The deployment method, or the use of Unicode filenames?
11:39 ndevos Alex: the deployment method looks ok, you tuned a lot, not sure if that is needed, but that is unrelated to the issue - I just never have seen anyone using unicode filenames on glusterfs
11:39 Alex Gotcha, thanks.
11:39 Alex Sorry, I'm only one-coffee-in still, so brain's somewhere between mush and working.
11:40 ndevos Alex: whay you could do, is try to use some other fuse filesystem and see if that has the same issue, if it has it may be a kernel/fuse problem
11:40 Alex Interestingly, a colleague of mine tried: "I've tried LVM to combine the disks on 1 box and do mirror and also there are no issues."
11:40 Alex which I think may rule out the fuse driver
11:41 Alex *not sure*. More testing to do when I get chance I guess
11:41 ndevos no, LVM is not using fuse, you can not compare that
11:41 Alex Ah, apologies - I assumed it would've still been mounted the same way.
11:42 ndevos oh, and in fact, building a volume with 6 bricks, where all the bricks are on the same server, does not make that much sense...
11:42 Alex Noted, in this case [I think] it's the first server of a multi server deployment.
11:42 ndevos no, LVM provides a block-device, it is more comparable to software-raid
11:43 ndevos that makes sense, gluster is very suitable for multiple servers, not so much for one server
11:47 ababu joined #gluster
11:47 dusmant joined #gluster
11:51 glusterbot New news from newglusterbugs: [Bug 1028672] BD xlator <http://goo.gl/y6DSIl> || [Bug 1028673] Add zerofill FOP to offload creating zeroed (VM) files <http://goo.gl/vbUwqx>
11:53 kanagaraj joined #gluster
12:01 RameshN joined #gluster
12:06 bala joined #gluster
12:43 bfoster joined #gluster
12:50 plarsen joined #gluster
13:10 satheesh joined #gluster
13:36 RicardoSSP joined #gluster
13:36 RicardoSSP joined #gluster
13:45 mjrosenb joined #gluster
13:50 hateya joined #gluster
13:54 kbut ndevos: thanks
14:19 ricky-ticky joined #gluster
14:26 ricky-ticky joined #gluster
14:34 hateya joined #gluster
14:42 ricky-ticky joined #gluster
14:50 hagarth joined #gluster
15:02 ricky-ticky joined #gluster
15:07 ricky-ticky joined #gluster
15:15 Rav_ joined #gluster
15:17 ricky-ticky joined #gluster
15:24 ricky-ticky joined #gluster
15:27 ricky-ticky1 joined #gluster
15:28 hateya joined #gluster
15:36 ricky-ticky joined #gluster
15:38 ricky-ticky1 joined #gluster
15:48 ricky-ticky joined #gluster
16:03 ricky-ticky joined #gluster
16:08 ricky-ticky joined #gluster
16:16 ricky-ticky joined #gluster
16:18 ricky-ticky1 joined #gluster
16:25 bulde joined #gluster
16:29 ricky-ticky joined #gluster
16:35 ricky-ticky joined #gluster
16:38 ricky-ticky1 joined #gluster
16:51 ricky-ticky joined #gluster
17:05 ricky-ticky joined #gluster
17:12 ricky-ticky joined #gluster
17:14 ricky-ticky1 joined #gluster
17:17 ricky-ticky joined #gluster
17:17 calum_ joined #gluster
17:19 ricky-ticky1 joined #gluster
17:20 ricky-ticky joined #gluster
17:21 Claer joined #gluster
17:21 Claer Hi
17:21 glusterbot Claer: 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.
17:22 ricky-ticky joined #gluster
17:24 Claer I'm trying to build my first shared FS with 2 nodes and 2 replicas for each file. OS are gentoo (gluster 3.4.0) and fedora (gluster 3.4.1)
17:24 Claer I successfully installed and started the daemons
17:26 social Claer: I wouldn't go with 3.4.0,  use latest stable
17:26 Claer 3.4.1 isnt packaged for gentoo
17:26 Claer I had no choice ;-)
17:26 social hmm shouldn't be hard for you to repack it even update it if you use gentoo it should be easy for you
17:27 Claer let me try
17:29 Claer while it is upgrading, I'll explain my issue
17:30 dbruhn joined #gluster
17:30 Claer After starting daemons, I can successully probe the peer to add it to the cluster
17:30 Claer then I try to create a volume and it fails with the following message
17:31 Claer volume create: shared_VMs: failed: Host <otherhost> is not in 'Peer in Cluster' state
17:31 social gluster peer status
17:32 Claer no error reported
17:32 social I would guess one of the peers got rejected
17:33 Claer I checked from both nodes. they both said the peer is connected
17:34 avati joined #gluster
17:36 social I would guess it is some kind of bug but not sure, there are some fixes that might be related in 3.4.1 that's why I would first ensure you are using that
17:37 Claer I have to update the 4 gentoo patches, I'll report in few mn
17:40 Remco There should be logfiles on both sides
17:41 ricky-ticky joined #gluster
17:47 Claer hum, I restarted the daemon on the gentoo side and it worked, strange.
17:47 Claer (I wanted to capture the exact logs)
17:48 ricky-ticky joined #gluster
17:51 ricky-ticky joined #gluster
17:53 ricky-ticky1 joined #gluster
18:03 ricky-ticky joined #gluster
18:05 ricky-ticky1 joined #gluster
18:07 ricky-ticky joined #gluster
18:17 ricky-ticky joined #gluster
18:19 jbrooks joined #gluster
18:19 ricky-ticky joined #gluster
18:22 ricky-ticky1 joined #gluster
18:22 eclectic_ joined #gluster
18:24 ricky-ticky joined #gluster
18:32 hateya joined #gluster
18:34 ricky-ticky joined #gluster
18:36 ricky-ticky joined #gluster
18:50 ricky-ticky joined #gluster
18:53 tommat joined #gluster
19:00 ricky-ticky joined #gluster
19:35 daMaestro joined #gluster
20:08 ada4u joined #gluster
20:08 ada4u Hi
20:08 glusterbot ada4u: 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.
20:15 ada4u Regards, little question about gluster and openvpn, my current config is host in network 1: a.a.a.a, vpn-server-tun: c.c.c.a and on network 2: b.b.b.b, vpn.client-tun: c.c.c.b, now from network a: ---> gluster peer probe c.c.c.b ; gluster volume create tst replica 2 transport tcp c.c.c.a:/mnt/disk c.c.c.b:/mnt/disk, local client on both servers can mount.glusterfs c.c.c.a:/tst and everithing...
20:15 ada4u ...works fine, but external client can't connect to a.a.a.a neither b.b.b.b. any hint on how to achieve connection whit external client... ?
21:05 Gugge you lost me at a.a.a.a
21:05 Gugge Try again, with the real adresses :)
21:19 elyograg http://www.portal2sounds.com/1900
21:19 glusterbot Title: Portal 2 Sounds | Wheatley: No. Wait, did I do B? Do you have a pen? Start writing these down. (at www.portal2sounds.com)
21:20 DV joined #gluster
22:10 badone joined #gluster
22:22 Claer cya
22:38 m0zes joined #gluster
22:48 m0zes joined #gluster

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