Camelia, the Perl 6 bug

IRC log for #gluster, 2013-05-27

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

All times shown according to UTC.

Time Nick Message
01:27 jclift_ joined #gluster
01:29 jclift joined #gluster
01:37 satheesh joined #gluster
01:58 lnxsix joined #gluster
01:58 nightwalk edoceo: nope, that won't accomplish *anything*. it's all-or-nothing
02:09 recidive_ joined #gluster
02:15 majeff joined #gluster
02:19 majeff left #gluster
02:21 wgao joined #gluster
02:22 kevein joined #gluster
02:56 lalatenduM joined #gluster
03:09 majeff joined #gluster
03:10 bharata joined #gluster
03:33 anands joined #gluster
03:45 mohankumar__ joined #gluster
03:46 majeff joined #gluster
03:53 _BryanHm_ joined #gluster
04:02 sgowda joined #gluster
04:06 atrius_ joined #gluster
04:15 bulde joined #gluster
04:20 vpshastry joined #gluster
04:25 ngoswami joined #gluster
04:50 aravindavk joined #gluster
05:01 saurabh joined #gluster
05:01 shylesh joined #gluster
05:08 hagarth joined #gluster
05:11 sprachgenerator joined #gluster
05:18 lalatenduM joined #gluster
05:21 lala_ joined #gluster
05:28 rastar joined #gluster
06:03 rgustafs joined #gluster
06:11 bala1 joined #gluster
06:12 psharma joined #gluster
06:24 vshankar joined #gluster
06:28 jtux joined #gluster
06:29 psharma joined #gluster
06:42 vimal joined #gluster
06:46 anands joined #gluster
06:46 kaushal joined #gluster
06:49 dobber_ joined #gluster
06:56 psharma joined #gluster
07:02 ctria joined #gluster
07:03 vpshastry1 joined #gluster
07:05 andreask joined #gluster
07:08 tjikkun_work joined #gluster
07:12 sgowda joined #gluster
07:14 anands joined #gluster
07:23 hybrid5121 joined #gluster
07:23 ujjain joined #gluster
07:36 puebele joined #gluster
07:39 sgowda joined #gluster
07:47 majeff joined #gluster
07:51 vpshastry1 joined #gluster
07:52 dobber_ joined #gluster
07:54 puebele joined #gluster
08:01 spider_fingers joined #gluster
08:02 psharma joined #gluster
08:09 tziOm joined #gluster
08:13 kelkoobenoitr joined #gluster
08:43 glusterbot New news from resolvedglusterbugs: [Bug 910239] open behind issue <http://goo.gl/i9ZRQ>
08:48 Rhomber it there a way to reduce the consistency level?
08:49 Rhomber i.e. I have 3 nodes, with replica 3.. if the file exists on the node being queried, i just want it to return .. i don't care if it's 'up to date'
08:49 Rhomber (in my case, the file will never change without having a new file name
08:59 bharata joined #gluster
09:05 kaushal_ joined #gluster
09:15 rb2k joined #gluster
09:20 majeff1 joined #gluster
09:24 hagarth joined #gluster
09:25 ricky-ticky joined #gluster
09:32 rastar joined #gluster
09:34 saurabh joined #gluster
09:44 lh joined #gluster
09:44 lh joined #gluster
09:53 kshlm joined #gluster
09:54 bleon joined #gluster
09:59 spider_fingers1 joined #gluster
10:06 majeff1 left #gluster
10:11 kris-- joined #gluster
10:12 edward1 joined #gluster
10:13 duerF joined #gluster
10:18 bharata joined #gluster
10:25 spider_fingers joined #gluster
10:27 vpshastry1 joined #gluster
10:36 piotrektt joined #gluster
10:42 hagarth joined #gluster
10:43 manik joined #gluster
10:51 spider_fingers left #gluster
11:09 manik joined #gluster
11:13 theron joined #gluster
11:16 vpshastry1 joined #gluster
11:19 ninkotech joined #gluster
11:19 ninkotech_ joined #gluster
11:41 krishna joined #gluster
12:02 ctria joined #gluster
12:04 ollivera joined #gluster
12:08 ollivera left #gluster
12:09 krishna__ joined #gluster
12:11 isomorphic joined #gluster
12:21 koubas joined #gluster
12:26 majeff joined #gluster
12:36 vpshastry1 joined #gluster
12:47 abyss^ It's possible to do smth like this: I have 2x VM (Xen Citrix) with glusterfs (server) and attached to this VMs disk via nfs where datas are lied. It's possible to set up new gluster servers and detached disk from old VMs and attach to new VMs and reconfigure clients and it's possible to work?
12:50 abyss^ Iam asking because when I copied some files to disk and make gluster volume and add the brick (folder with datas) client can't see this data, I had to copy this files via client...
12:51 recidive_ joined #gluster
12:51 abyss^ only then clients can see data...
12:57 abyss^ If I wrote smth unclear I can clarify that :)
13:07 puebele3 joined #gluster
13:08 krishna__ joined #gluster
13:14 koubas joined #gluster
13:17 16WAADQW1 joined #gluster
13:18 krishna joined #gluster
13:20 ctria joined #gluster
13:24 krishna__ joined #gluster
13:31 vpshastry1 joined #gluster
13:33 lbalbalba joined #gluster
13:37 puebele1 joined #gluster
13:41 StarBeast joined #gluster
13:47 krishna__ joined #gluster
13:51 krishna__ joined #gluster
13:56 puebele1 joined #gluster
13:57 krishna joined #gluster
14:01 mohankumar__ joined #gluster
14:06 krishna__ joined #gluster
14:10 lbalbalba hi. im running into an issue when trying to compile samba-glusterfs-vfs. make complains that it cant find config.h in some dir. im wondering, is there a recommended samba version to use when trying to build this ? im using latest git of samba
14:11 45PAAGDME joined #gluster
14:12 jiku joined #gluster
14:12 jikz_ joined #gluster
14:14 piotrektt joined #gluster
14:16 NuxRo lbalbalba: I would imagine you need samba-devel installed?
14:17 NuxRo or/and glusterfs-devel
14:18 lbalbalba NuxRo: thx, i compiled from sources, for both samba4 and glusterfs. i did run ./configure and make, so that config.h should just be there, right ?
14:18 NuxRo I think you're supposed to use samba3
14:18 lbalbalba ah. that explains a lot.
14:19 lbalbalba thx
14:19 lbalbalba ill try that
14:19 NuxRo np, good luck, let us know if it works well
14:33 charlescooke joined #gluster
14:39 ndevos lbalbalba: the samba-glusterfs-vfs from forge.gluster.org looks a little empty, you should rather try http://www.mail-archive.com/glust​er-devel@nongnu.org/msg09514.html
14:39 glusterbot <http://goo.gl/2AwRa> (at www.mail-archive.com)
14:39 jtux joined #gluster
14:41 lbalbalba ndevos: thx. weird though, it 'should' just be in the git repo ;)
14:42 lbalbalba ndevos: is there an eazy/lazy way to cut that page up into the multiple files ?
14:42 ndevos lbalbalba: thats what I hoped too, but src/vfs_glusterfs.c does not contain any functionality
14:44 ndevos lbalbalba: if you save the raw email you can use 'git am' to apply it to the samba sources
14:44 ndevos lbalbalba: if you dont have the email, you can use http://paste.fedoraproject.org/14773/66581213 :)
14:44 glusterbot Title: #14773 Fedora Project Pastebin (at paste.fedoraproject.org)
14:45 ndevos (use the 'raw' version and save that as a file, and use 'git am $file' in the samba git repo)
14:46 krishna__ joined #gluster
14:48 vpshastry joined #gluster
14:48 vpshastry left #gluster
14:49 lbalbalba ndevos: http://fpaste.org/14777/36966614/
14:49 glusterbot Title: #14777 Fedora Project Pastebin (at fpaste.org)
14:50 ndevos lbalbalba: to what repository are you applying that patch with 'git am'?
14:50 lbalbalba ndevos: oh. i need to this in the samba3 source dir.
14:50 ndevos lbalbalba: it should be the one from git://git.samba.org/samba.git (it applies correctly to the master branch)
14:53 lbalbalba ah. patch the samab git repo. i assumed it was a build 'outsode' of the samba source code tree. my bad
14:53 lbalbalba samab/samba
14:53 lbalbalba outsod/outside
14:54 lbalbalba :(
14:57 lbalbalba ah. that works : Applying: vfs_glusterfs: Samba VFS module for glusterfs
15:01 lbalbalba thanks
15:02 ndevos lbalbalba: how have you installed glusterfs? you will need a /usr/lib(64)/pkgconfig/glusterfs-api.pc file for compiling the vfs_glusters module
15:02 * ndevos just hit that issue :-/
15:03 lbalbalba ndevos: ./configure && make && make install
15:06 lbalbalba ndevos: i have '/usr/local/glusterfs/lib/p​kgconfig/glusterfs-api.pc'
15:06 ndevos lbalbalba: okay, that may work too :)
15:06 lbalbalba ndevos: i prefixed configure '/usr/local/glusterfs/'.
15:09 ndevos lbalbalba: sure, that should be fine, I'm trying with RPMs on Fedora 18 and thats missing the .pc file :-/
15:12 clag_ joined #gluster
15:24 andreask joined #gluster
15:33 krishna__ joined #gluster
15:37 lbalbalba patch aplies, configure runs, make succeeded ! now 'all' thats left is for me to figure out how samba works :P
15:38 lbalbalba where's the quickstart for samba vfs ?
15:42 hagarth joined #gluster
15:53 ndevos lbalbalba: like this: http://lists.nongnu.org/archive/html​/gluster-devel/2013-04/msg00098.html
15:53 glusterbot <http://goo.gl/KFgT8> (at lists.nongnu.org)
15:54 leaky joined #gluster
15:54 leaky hello guys, i'm having trouble with xattr fuse under ubuntu 12.04 - starting ctdb / samba cluster...
15:55 leaky anyone else had issues with extended attributes on their gluster mounts?
15:56 ndevos leaky: the only thing that really matters (that I am aware of) is to use big inodes, like 'mkfs -t xfs -i size=512 ...', or even '-i size=1024' - I'm not aware of issues
15:57 lbalbalba ndevos: thanks.
15:59 leaky wow, thats interesting... I'm using zfs underneath, maybe I should reformat (with large inodes?)
16:11 krishna__ joined #gluster
16:17 awheeler_ joined #gluster
16:31 hybrid5121 joined #gluster
16:55 anands joined #gluster
17:00 Maskull joined #gluster
17:01 Maskull heya guys, i've a question regarding distributed file systems in general
17:02 Maskull are there any advantages/ disadvantages using glusterfs over nfs or ceph or so?
17:05 NuxRo Maskull: re glusterfs vs nfs, the biggest selling point for glusterfs is out of the box HA and horizontal scalability
17:05 NuxRo can't speak for ceph, don't use it, i imagine it has similar features
17:09 Maskull In general speaking ceph and glusterfs have the features, so there's not a specific reason why to use one over the other?
17:10 Maskull from what ive heard so far is that nfs is quite different than gluster
17:16 NuxRo glusterfs comes with a built-in nfsv3 server, so you can use that
17:16 NuxRo but yes, nfs and glusterfs are different
17:17 lbalbalba Maskull: As far as i know, glustrfs has client side failover, and nfs doesnt
17:18 lbalbalba Maskull: NFS: server fails: fsckd  :glusterfs fails: on to next node/brick
17:18 Maskull yes, that's one of my main concerns aswell about NFS having a single point of failure
17:22 anands joined #gluster
17:27 js_ joined #gluster
17:28 js_ using glusterfs, will every connected node add to HA?
17:28 JoeJulian js_: No, only servers.
17:28 joelwallis joined #gluster
17:28 js_ ah, right
17:28 JoeJulian And then, only replicas.
17:30 js_ do the servers typically need a lot of ram?
17:34 lanning joined #gluster
17:34 zaitcev joined #gluster
17:35 JoeJulian Depends on how many bricks per server. I run 60 bricks in 16gig.
17:35 meunierd1 joined #gluster
17:36 vpshastry joined #gluster
17:37 js_ JoeJulian: sounds like quite a lot? i'm just looking for HA shared storage for one volume, so i guess one brick then
17:44 vpshastry left #gluster
17:50 ska joined #gluster
17:51 ska I'd like to make a cluster filesystem with only one server (for now) that can be RW mounted on multiple clients. Is there a howto on such a thing/
17:51 ska ?
18:00 saurabh joined #gluster
18:00 _BryanHm_ left #gluster
18:02 ninkotech ska: yes :)
18:02 ninkotech ska: did you try your fav search engine?  glusterfs single server installation  howto ?
18:03 ska ok.. ty
18:04 ninkotech ska: i dont really know one from my head, but there definitelly are some
18:06 krishna__ joined #gluster
18:09 hagarth ska: http://www.howtoforge.com/creating-an​-nfs-like-standalone-storage-server-w​ith-glusterfs-3.2.x-on-ubuntu-12.04
18:09 glusterbot <http://goo.gl/tJkVi> (at www.howtoforge.com)
18:12 ninkotech hagarth: ty
18:12 ctria joined #gluster
18:15 meunierd1 Hi all, I'm having difficulty setting up a trusted storage pool between EC2 instances. I run "> peer probe subdomain.domain.ca" and after a few minutes there's no output and the prompt is returned, then after running "> peer status" I get 'Status: Establishing Connection (Disconnected)'
18:16 krishna__ left #gluster
18:16 ska Are there Glusterfs rpms for SLES 11 SP2?
18:18 meunierd1 Whoops, I was just forgetting to 'Apply Rule Changes' from the AWS dash. Disregard :)
18:20 ska I'm worried that my newere server may have a version of glusterfs that is way newer than the old RPM's I may find for the clients.
18:36 jag3773 joined #gluster
18:40 failshell joined #gluster
18:40 failshell how does gluster handle bricks that don't have the same storage capacity?
18:40 failshell is it going to store more on the one with more storage?
18:49 StarBeast joined #gluster
18:58 leaky failshell: are you refering to replica volumes?
19:02 leaky if you create a volume with replica count 2, it will ensure there are two copies of all files - that is all...
19:03 leaky if you have one brick larger than the other, you should be limited by the smaller brick - if i'm not mistaken
19:05 failshell ok, so ill keep adding 128GB per brick then
19:16 lbalbalba ska: im not sure, but i think that (up untill now, at least) its recommended to use the same version for all client and servers. YMMV
19:20 lbalbalba @rpm
19:20 glusterbot lbalbalba: I do not know about 'rpm', but I do know about these similar topics: 'rdma', 'rtfm'
19:21 lbalbalba crap
19:21 JoeJulian @yum
19:21 glusterbot JoeJulian: The official glusterfs packages for RHEL/CentOS/SL are available here: http://goo.gl/s077x
19:24 JoeJulian failshell: See my article "dht misses are expensive" which explains how the dht algorithm works. From that you can understand what's going to happen when dht subvolume sizes are different. Short answer, when the smaller subvolume is full, it becomes inefficient.
19:24 JoeJulian @lucky dht misses are expensive
19:24 glusterbot JoeJulian: http://goo.gl/A3mCk
19:26 rotbeard joined #gluster
19:27 lbalbalba @yum | ska
19:27 failshell JoeJulian: ok, im gonna have to live with that 128GB then
19:28 failshell and have more VMs
19:33 meunierd1 If I want to replicate a brick, should it be empty before I mount it?
19:34 meunierd1 ie should I only manipulate the data through the mounted gluster volume or can it be prepopulated with some data?
19:41 failshell dont take my word for it, but  i only manipulate them via a mounted volume
19:43 hjmangalam1 joined #gluster
19:44 lh joined #gluster
19:53 meunierd1 Also, I don't understand how to create a volume between two servers if I can't specify localhost from the server I'm running the command on
19:53 edoceo it's recommened to only access via the mount point; but I've created a brick that points to an existing directory when using Gluster like a single NFS server style
19:53 edoceo Then lateter you can add a brick to make a replica
19:54 edoceo But that new brick starts out with an empty directory, and you only access that via the mount point; and you may have to say "rebalance" and do the `find` trick
19:56 andreask joined #gluster
19:58 ska @yum
19:58 glusterbot ska: The official glusterfs packages for RHEL/CentOS/SL are available here: http://goo.gl/s077x
19:58 meunierd1 but how do I refer to a local export directory?
20:06 ctria edoceo, i don't think what you state would work...
20:06 ctria you should not access the brick mount point directly
20:07 ctria this is expected to not work as you expect.
20:07 ctria :)
20:07 edoceo I was saing to *not* access directly
20:07 ctria ah ok i misread it then :)
20:08 edoceo i was saying that you could create a brick pointing to an existing, populated directory for one GlusterFS, but then on only access via GlusterFS mount
20:11 meunierd1 If I try to reference the brick by external ip, I'm told it's not a friend
20:11 meunierd1 if I use the internal IP the operation fails nondescriptly
20:16 edoceo in what operation? adding a peer?  mounting?
20:16 meunierd1 creating a volume
20:17 meunierd1 I have glusterfs installed on two EC2 instances, each with a brick.
20:17 meunierd1 I want to create a replica volume between them
20:17 meunierd1 I was able to successfully add the second server as a peer.
20:19 meunierd1 so then from the quickstart guide it says:
20:20 meunierd1 node01.mydomain.net:/export/brick1/sdb1
20:20 meunierd1 I don't really understand how to refer to node01
20:22 meunierd1 If I try the domain name of the instance, it says 'not a friend'
20:25 GLHMarmot joined #gluster
20:38 meunierd1 http://hastebin.com/qaratocobe.sm -- this is the gluster log of the second server
20:38 glusterbot Title: hastebin (at hastebin.com)
20:39 meunierd1 whoops some was cut off: http://hastebin.com/guguyimapa.sm
20:39 glusterbot Title: hastebin (at hastebin.com)
20:46 manik joined #gluster
21:17 edoceo !seen meunierd1
21:17 edoceo oh well
21:27 elyograg joined #gluster
21:27 elyograg left #gluster
21:31 duerF joined #gluster
21:32 theron joined #gluster
21:35 vex is the process for removing bricks from a volume similar to adding? I can remove a brick and the data stays intact?
21:36 rb2k joined #gluster
22:09 yinyin joined #gluster
22:11 leaky joined #gluster
22:20 leaky left #gluster
23:01 ccha joined #gluster
23:03 isomorphic joined #gluster
23:18 edoceo vex: the brick will be removed the the Gluster pool but the data on the filesystem will remain
23:21 vex cool. that's what I thought. thanks.
23:22 vex wait. I mean data in the gluster volume, not on the filesystem on the disk
23:24 yinyin joined #gluster
23:52 edoceo If you have only one Brick in the volume and you remove that brick there will be no data in the volume; if you have multiple bricks with replicate (or disttribute/replicate) then your data should remain

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