Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster, 2014-10-08

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

All times shown according to UTC.

Time Nick Message
00:28 justinmburrous joined #gluster
00:53 rotbeard joined #gluster
00:53 quique left #gluster
00:57 uebera|| joined #gluster
00:57 uebera|| joined #gluster
00:57 cjanbanan joined #gluster
01:03 rotbeard joined #gluster
01:04 uebera|| joined #gluster
01:07 NuxRo joined #gluster
01:09 stickyboy joined #gluster
01:12 uebera|| joined #gluster
01:12 uebera|| joined #gluster
01:17 uebera|| joined #gluster
01:19 rotbeard joined #gluster
01:22 jmarley joined #gluster
01:30 marcoceppi_ joined #gluster
01:32 plarsen joined #gluster
01:36 justinmburrous joined #gluster
01:57 wgao joined #gluster
02:15 cjanbanan joined #gluster
02:26 lyang0 joined #gluster
02:29 rafi joined #gluster
02:34 justinmburrous joined #gluster
02:41 bharata-rao joined #gluster
02:46 overclk joined #gluster
02:50 overclk joined #gluster
02:58 cjanbanan joined #gluster
03:07 toordog joined #gluster
03:31 bala joined #gluster
03:42 itisravi joined #gluster
03:43 shubhendu joined #gluster
03:53 nbalachandran joined #gluster
03:56 kshlm joined #gluster
04:00 doubt joined #gluster
04:29 rafi joined #gluster
04:29 Rafi_kc joined #gluster
04:30 cjanbanan joined #gluster
04:40 spandit joined #gluster
04:44 jiffin joined #gluster
04:49 ramteid joined #gluster
04:58 bala joined #gluster
04:59 lalatenduM joined #gluster
05:04 smohan joined #gluster
05:13 shubhendu joined #gluster
05:13 ndarshan joined #gluster
05:18 smohan joined #gluster
05:19 kdhananjay joined #gluster
05:21 smohan_ joined #gluster
05:25 overclk joined #gluster
05:26 kanagaraj joined #gluster
05:29 smohan joined #gluster
05:30 smohan_ joined #gluster
05:30 cjanbanan joined #gluster
05:31 chirino joined #gluster
05:31 prasanth_ joined #gluster
05:33 smohan_ joined #gluster
05:33 elico joined #gluster
05:34 raghu joined #gluster
05:39 Pupeno joined #gluster
05:47 atalur joined #gluster
05:49 dusmant joined #gluster
05:49 panpanfeng joined #gluster
06:03 overclk joined #gluster
06:03 atalur joined #gluster
06:09 LebedevRI joined #gluster
06:13 Kins joined #gluster
06:28 Philambdo joined #gluster
06:32 nishanth joined #gluster
06:34 raghu` joined #gluster
06:39 RaSTar joined #gluster
06:43 ctria joined #gluster
06:46 saurabh joined #gluster
06:59 ekuric joined #gluster
07:02 justinmburrous joined #gluster
07:02 DV joined #gluster
07:03 cjanbanan joined #gluster
07:09 Fen2 joined #gluster
07:10 ACiDGRiM joined #gluster
07:11 ACiDGRiM If anyone is active I'm trying to build gluster on solaris 11 to take advantage of ZFS encryption but since glusterfs seems to be more linux targed I'm having a bad time
07:12 ACiDGRiM Is this a common thing? my searching hasn't gotten me past gmake throwing all-recursive errors
07:16 soumya_ joined #gluster
07:17 cjanbanan joined #gluster
07:18 Slashman joined #gluster
07:21 itisravi joined #gluster
07:28 codex joined #gluster
07:29 fsimonce joined #gluster
07:31 ronis joined #gluster
07:38 cjanbanan joined #gluster
07:43 Fen2 Not common, RHEL/CentOS/Ubuntu/Fedora are recommended, not solaris
07:45 Fen2 But i found this in the doc : "Windows, Mac, Unix, Linux and other Unix-like operating systems are all supported"
07:54 Fen2 nobody use the gluster management console ?
07:55 rjoseph joined #gluster
07:56 sickness yeah, I've tried to search info on the net but little arised
07:56 sickness just people asking the same thing on some mls
07:56 sickness I was curious to try to run the server on cygwin :p
08:04 rolfb joined #gluster
08:11 ronis joined #gluster
08:14 aravindavk joined #gluster
08:15 anands joined #gluster
08:18 ws2k3 Fen2 what exacly do you mean with gluster management console ?
08:18 Fen2 The graphic interface
08:19 ws2k3 inside ovirt or ?
08:19 ws2k3 cause there used to be an web interface for gluster but that a long time ago that is not supported anymore
08:26 ws2k3 anyone has some information about monitoring glusterfs using nagios? possible without nrpe?
08:28 itisravi joined #gluster
08:31 justinmburrous joined #gluster
08:35 mrEriksson ws2k3: What do you want to monitor? I just make sure that the ports are up and that the processes are running (via snmp). That does not collect statistics for me though
08:37 johndescs currently found that : https://github.com/pcuzner/gluster-monitor
08:37 glusterbot Title: pcuzner/gluster-monitor · GitHub (at github.com)
08:38 johndescs not sure how to plug it into nagios / whatever snmp monitoring
08:41 ws2k3 johndescs i would like to know howmany brick are online then it notify's me when something goes wrong
08:43 johndescs me too :)
08:44 ws2k3 and if possible i wanne do it without nrpe i dont wanne install any nagios agents stuff on my gluster server
08:44 ws2k3 my gluster server should do what it supposed to do and that is being a gluster server :)
08:44 Fen2 So this doesn't exist anymore ? http://gluster.org/community/documentation/images/2/2f/Gsp-31-console-exploring.png
08:45 ws2k3 Fen2 no that doesnt exsist anymore howevery what you could try is install ovirt
08:45 ws2k3 ovirt still has a web interface for glusterfs
08:45 Fen2 we are using proxmox
08:45 ws2k3 yeah i know but you can maby make a virtual machine or something and install ovirt there
08:46 ws2k3 and only use ovirt for managing glusterfs you dont have to use ovirt for virtualisation if you dont want to
08:46 ws2k3 at least you reach your goal then by having a web interface for glusterfs
08:47 Fen2 hum ok sounds good :p
08:48 ws2k3 http://www.ovirt.org/images/5/51/Georepsession1list.png
08:48 dusmant joined #gluster
08:49 ws2k3 fen2 let me know if you wanne give it a try cause i'm intrested in using ovirt to however i never used/installed it myself still need to look into that so let me know if it is any good
08:49 Fen2 i will certainly try but not immediatly :)
08:50 ws2k3 sure np :) let me know what you think about it i would appriciate that :)
08:50 ndarshan joined #gluster
08:50 nishanth joined #gluster
08:52 panpanfeng hello , anyone can help me ?  I met a problem. when I use gluster 3.6 client mount gluster 3.3 volume , the write operation will fail.
08:54 ws2k3 panpanfeng i think you can already guess... 99 % sure its a compatebility issue
08:54 ws2k3 try to use the same client and server version
08:55 Fen2 3.6 ? just a beta version, maybe it can be also your problem
08:55 panpanfeng there are servers with gluster 3.3 version  which are running in production envrionment.
08:56 panpanfeng upgrading the version to 3.6 may cause some safe problem.
08:57 panpanfeng Fen2  yeah, it's still a beta version , but i use this version for ec translator
08:57 panpanfeng which my definitely reduce the storage usage.
08:58 panpanfeng I have test gluster 3.6 client and gluster 3.4 server , under this condition, everything seems OK.
08:59 panpanfeng Has anybody met this compatebility issue ?
09:01 dusmant joined #gluster
09:10 WvD joined #gluster
09:11 ndarshan joined #gluster
09:12 vimal joined #gluster
09:12 nishanth joined #gluster
09:13 WvD_dev Hi there
09:13 WvD_dev Quick question: How can I inspect the lock tables from the linux command-line?
09:13 WvD_dev I'm using GlusterFS 3.2
09:17 prasanth_ joined #gluster
09:18 dusmant joined #gluster
09:19 Fen2 ws2k
09:20 Fen2 ws2k3: i will maybe use Nagios GUI for monitoring. And what is the diference between red hat storage and glusterfs ?
09:23 ws2k3 to be honest i dont know but i do know gluster is managed by red hat so i think redhat storege server is just a linux distro with glusterfs build in maby a nice web interface so its kinda the enterprice server of gluster with red hat support
09:28 harish joined #gluster
09:28 Fen2 ok thx :)
09:29 Fen2 and even if the glusterfs GUI is no more supported, can we still install it ? (with gluster 3.5.2)
09:30 ws2k3 as far as i know not
09:30 ws2k3 the web gui was from the time from gluster 2.* so probarly thats not gonna work
09:30 ws2k3 if you want a web interface try ovirt that the only way as far as i know
09:31 kdhananjay1 joined #gluster
09:43 panpanfeng hello , anyone can help me ?  I met a problem. when I use gluster 3.6 client mount gluster 3.3 volume , the write operation will fail.
09:48 haomaiwa_ joined #gluster
09:54 nbalachandran joined #gluster
10:03 morse joined #gluster
10:08 DV joined #gluster
10:16 Pupeno_ joined #gluster
10:18 justinmburrous joined #gluster
10:34 nullck joined #gluster
10:42 bharata_ joined #gluster
10:46 DV joined #gluster
10:49 diegows joined #gluster
10:53 Philambdo joined #gluster
10:55 the-me joined #gluster
10:57 nbalachandran joined #gluster
10:58 kdhananjay joined #gluster
11:00 ira joined #gluster
11:00 prasanth_ joined #gluster
11:06 justinmburrous joined #gluster
11:11 eryc joined #gluster
11:11 partner joined #gluster
11:11 urban joined #gluster
11:11 dockbram joined #gluster
11:11 al joined #gluster
11:11 morsik joined #gluster
11:11 ultrabizweb joined #gluster
11:11 haakon_ joined #gluster
11:11 msvbhat joined #gluster
11:11 DV joined #gluster
11:11 msvbhat joined #gluster
11:11 soumya_ joined #gluster
11:11 JordanHackworth joined #gluster
11:11 soumya_ joined #gluster
11:11 eightyeight joined #gluster
11:11 Pupeno joined #gluster
11:11 osiekhan1 joined #gluster
11:11 tom[] joined #gluster
11:11 frb joined #gluster
11:12 T0aD joined #gluster
11:12 kalzz joined #gluster
11:12 asku joined #gluster
11:12 mojibake joined #gluster
11:13 virusuy joined #gluster
11:16 foster joined #gluster
11:16 johnnytran joined #gluster
11:19 cmtime joined #gluster
11:29 overclk joined #gluster
11:39 jmarley joined #gluster
11:47 soumya joined #gluster
11:53 Fen1 joined #gluster
11:55 bene2 joined #gluster
11:55 bene2 is there a meeting right now for the community?
11:55 RaSTar joined #gluster
11:56 jdarcy joined #gluster
11:58 Fen1 yeah but you need to come here, in france :p
11:59 Fen1 http://blog.gluster.org/meetups/
11:59 prasanth_ joined #gluster
12:02 Slashman joined #gluster
12:03 justinmburrous joined #gluster
12:04 anands joined #gluster
12:16 Fen1 bene2: Or go on the #gluster-meeting the 12th october
12:16 Fen1 sry 14th
12:17 bene2 thx
12:18 bene2 Fen1, would be happy to come there!
12:25 ppai joined #gluster
12:27 calum_ joined #gluster
12:28 tdasilva joined #gluster
12:29 calisto joined #gluster
12:30 justinmburrous joined #gluster
12:30 ws2k3 fwn1 where ?
12:30 ws2k3 Fen1 where ?
12:30 julim joined #gluster
12:31 Fen1 ws2k3, in france
12:32 Fen1 near the famous museum
12:32 Fen1 at Paris
12:32 ws2k3 ah oke a bit to far for me :)
12:36 soumya joined #gluster
12:38 calisto hi all people en this channel, I have a doubt... I create a laboratory, 2 machines node1 and node2 in node 1 run mount -t glusterfs node1:/testvol /mnt/node1 and mount -t glusterfs node2:testvol /mnt/node2, /mnt are  have a file pr size 1GB the command time cat pr > /dev/null  produce  the same time runs on /mnt/node1 and /mnt/node2 and not have net trafic  on the node2, how is possible this? thanks
12:40 shubhendu joined #gluster
12:41 Fen1 which one is your server and client ?
12:42 Fen1 why do you mount node1 on node1 ?
12:42 bennyturns joined #gluster
12:42 Fen1 and node2 on node2 ?
12:43 Fen1 where is your volume ?
12:43 Fen1 calisto: ?
12:43 bala joined #gluster
12:44 calisto Fen1: In node1 I need mount the node 1 to work in a virtual machine in this computer
12:45 calisto Fen1 I explain a simplicity scheme
12:46 calisto supose I have a ftp server on node1
12:46 Fen1 you have gluster-server and your Hypervisor on the same serveur ?
12:46 calisto and I need replica of data on node2
12:47 calisto If any problem on node1 (suppose hardware probem) I need tu up the ftp server on node2
12:47 calisto yes Fen1
12:48 calisto I confgure the ftp server on node1 and mount using fuse the file sistem of gluster of the node1
12:48 Fen1 why you use glusterfs so ? It's a distributed solution, but if you use your client & server on the same machine, it's useless
12:49 calisto and create a replica on the brick of node2
12:49 Fen1 you need 3 machines...
12:49 Fen1 2 server and 1 client
12:49 calisto why
12:49 Fen1 because :p
12:50 Fen1 it's the way glusterfs works
12:50 Fen1 separating server (data) and client
12:51 Fen1 if you want a good laboratory, create 3 VM
12:51 Fen1 install 2 gluster-serveur and 1 gluster-client
12:53 Fen1 create a pool with your 2 server (to put your storage together) and distibute it to your client
12:54 Fen1 calisto: because now node1 is what... server or client ? and node2 ?
12:54 saurabh joined #gluster
12:54 anands joined #gluster
12:54 calisto ok Fen1, but I dont see a technical problem in make a ftp server  in the same mahine has the ftp server, whats the problem in this?
12:55 Fen1 calisto : a ftp server with a ftp server ?
12:57 calisto simply, make a volume test over the node1 and node2 as replica, and mount in node1 mount -t glusterfs 127.0.0.1/test /mnt/ftpdata and configure a proftp to server data on the firectory /mnt/ftpdata, and works very fine
12:59 Fen1 Ok i see... it's  really dirty
12:59 Fen1 node 1 = server + client (with FTP to access file)
13:00 Fen1 same for the node2
13:00 Fen1 is it right ?
13:00 Fen1 and replica between node1-server and node2-server
13:01 Fen1 calisto: is it that ?
13:02 calisto yes but Idont see te problem why not dirty nfs or cifs over gluster and is dirty ftp?? I dont understand
13:02 theron joined #gluster
13:02 Fen1 it's dirty with fuse/nfs/cifs/ftp :p
13:03 Fen1 it's dirty to install server and client on the same machine
13:03 aravindavk joined #gluster
13:03 jmarley joined #gluster
13:03 coredump joined #gluster
13:04 calisto the replica in node2 is the "real time backup" is any problem on the node1 I mount the node2 directory, add a IP on node 2 and  start ftp servervice onnode 2 and any works fine
13:04 Fen1 just use DRBD or an other mirror thing but not glusterfs to do that :p
13:04 justinmburrous joined #gluster
13:05 Fen1 calisto: so what is your probleme if it work fine ?
13:05 calisto Fen1 ok its dirty why in the documentation of gluster propose use gluster has NFS server...?
13:06 calisto the problem is I explain when start the comunication and you no focus in this... jajaja I explain
13:06 Fen1 gluster = nfs server ???!!!
13:06 calisto its a doubt not a problem
13:07 calisto gluster <> nfs but in the documentation explain you can use gluster as NFS/CIfs server
13:07 calisto OK  explain de doubt:
13:08 Rafi_kc left #gluster
13:09 Fen1 You can use FUSE/NFS/CIFS to access data : https://access.redhat.com/documentation/en-US/Red_Hat_Storage/3/html/Administration_Guide/sect-NFS.html
13:09 glusterbot Title: 7.3. NFS (at access.redhat.com)
13:09 calisto Have 2 machines node1 and node2 in node 1 mount -t glusterfs node1:/testvol /mnt/node1 and mount -t glusterfs node2:testvol /mnt/node2 in /mnt/node1 or /mnt/node2 (its the same)   have a file "pr" size 1GB the command "time cat pr > /dev/null"  produce  the same time runs on /mnt/node1 and /mnt/node2 and not have net trafic  on the node2, how is possible this? thanks
13:10 ws2k3 just finished setting up monitoring glusterfs with nagios if anyone wanne know how let me know
13:11 calisto Yes Fen1 I know its posible acces data usingo Fuse,NFS or cifs I see any tutos, this not a problem
13:11 Fen1 ws2k3: maybe in fews week thx :p
13:11 Fen1 calisto : let me undestand...
13:11 sprachgenerator joined #gluster
13:12 Fen1 I sum up :
13:12 h0cin joined #gluster
13:12 Fen1 Node 1 =>  mount -t glusterfs node1:/testvol /mnt/node1
13:13 calisto yes
13:13 Fen1 Node 1 => mount -t glusterfs node2:testvol /mnt/node2
13:13 Fen1 or node 2 ?
13:14 calisto Node 1 => mount -t glusterfs node2:testvol /mnt/node2 <--- this is correcto
13:14 glusterbot calisto: <-'s karma is now -2
13:14 Fen1 xD
13:14 smohan joined #gluster
13:14 nbalachandran joined #gluster
13:14 Fen1 calisto : ok
13:14 dusmant joined #gluster
13:15 Fen1 if your "testvol" is replicated between node1 and node2 you don't need to mount twice
13:16 calisto yes I dont need
13:16 calisto but is a test
13:16 Fen1 ok you want to access by the 2 node
13:16 rjoseph joined #gluster
13:17 calisto cat /mnt/node1/pr <-- dont create net traphic  and cat /mnt/node2/pr yes but in the reality 2 cat not perform net traphic and this is the cuestion, why?
13:17 glusterbot calisto: <'s karma is now -5
13:18 Fen1 i guess glusterfs choose a way to access volume, and if this way crash, it take the other way, no matter you write node2:/testvol
13:18 tdasilva joined #gluster
13:18 calisto why glusterbot decrement my kerma???
13:18 Fen1 don't know
13:18 rwheeler joined #gluster
13:19 Fen1 calisto : why glusterfs would take the longest way :p
13:21 calisto its possible
13:21 mojibake Can someone guide to howto on setting up a iocache volume, I thought I remember seeing in some documentation somewhere. Was reminded of it when testing SoftNAS and setting read cache in SoftNAS. Seems would get a performance boost on EC2 instance if set on the instance store.
13:23 calisto I dont know about this mojibake sorry
13:23 overclk_ joined #gluster
13:23 Fen1 mojibake: sry can't help for that :/
13:24 anands joined #gluster
13:26 ppai joined #gluster
13:27 mojibake Maybe I was thinking of io-cache. http://www.gluster.org/community/documentation/index.php/Documenting_the_undocumented but I thought there some some type of cache volume that could be created.
13:29 calisto Fen1: its possible force  the mount to mount the export  volume in specific server and not in another?
13:32 Fen1 calisto: hum... you want to force gluster to take the long way ?
13:32 calisto yes
13:33 calisto its a test
13:34 msmith_ joined #gluster
13:34 Fen1 calisto : i'm not an expert but i don't think, because your volume is an entity and gluster is optimised to access as faster as it can to this entity
13:34 justinmburrous joined #gluster
13:35 Fen1 calisto : but maybe if your main network is busy, glusterfs will choose the other way
13:36 Fen1 calisto : look at the option when you mount it
13:38 calisto ok
13:38 calisto thanks Fen1
13:39 Fen1 calisto : de nada
13:41 haomaiwang joined #gluster
13:51 overclk joined #gluster
13:53 Guest533 joined #gluster
13:59 bala joined #gluster
14:03 bene joined #gluster
14:03 SmithyUK I'm seeing "failed to get trusted.distribute.linkinfo key - Success" spammed in my rebalance logs and related bug ticket is closed? v3.5.2
14:03 SmithyUK https://bugzilla.redhat.com/show_bug.cgi?id=1037515
14:03 glusterbot Bug 1037515: high, high, RHS 2.1.2, pkarampu, CLOSED ERRATA, BVT: " failed to get trusted.distribute.linkinfo key" errors are seen in rebalance logs
14:04 SmithyUK Any idea what that might be related to?
14:13 lalatenduM SmithyUK, according to https://bugzilla.redhat.com/show_bug.cgi?id=1040356 , it was fixed in 3.5.0
14:13 glusterbot Bug 1040356: high, high, ---, pkarampu, CLOSED CURRENTRELEASE, BVT: " failed to get trusted.distribute.linkinfo key" errors are seen in rebalance logs
14:16 Pupeno_ joined #gluster
14:19 jobewan joined #gluster
14:24 SmithyUK $ rpm -q glusterfs  glusterfs-3.5.2-1.el5
14:25 SmithyUK It doesn't seem to be moving any data either
14:26 SmithyUK Does it do some sort of integrity checking before attempting to move data or should it just kick straight off?
14:26 SmithyUK Should add that this is a distributed volume
14:26 ntt joined #gluster
14:28 wushudoin joined #gluster
14:35 haomaiwa_ joined #gluster
14:47 justinmburrous joined #gluster
14:54 tdasilva joined #gluster
14:55 kshlm joined #gluster
14:58 ricky-ticky joined #gluster
15:01 Fen1 is it wrong to use the same partition for different volume ?
15:04 giannello joined #gluster
15:05 aravindavk joined #gluster
15:08 ronis_ joined #gluster
15:32 R0ok_ joined #gluster
15:34 kdhananjay joined #gluster
15:35 sprachgenerator joined #gluster
15:35 soumya joined #gluster
15:39 daMaestro joined #gluster
15:39 rotbeard joined #gluster
15:40 sputnik13 joined #gluster
15:54 ghenry joined #gluster
15:54 ghenry joined #gluster
15:54 lmickh joined #gluster
15:57 SteveCooling joined #gluster
15:59 SmithyUK Managed to get my rebalance working after seeing the issue I posted above. One of my gluster servers had an old daemon running alongside an updated one somehow, seemed to be messing up a lot of stuff
15:59 ntt Hi, should i restart gluster after doing setfattr? i have "Extended attribute trusted.glusterfs.volume-id is absent" error in my log
16:02 julim joined #gluster
16:07 dtrainor joined #gluster
16:16 discernica joined #gluster
16:17 justinmburrous joined #gluster
16:17 ACiDGRiM joined #gluster
16:19 cjanbanan joined #gluster
16:24 discernica Has anyone encountered a seemingly wrong disk block size issue? I am trying to install a commercial app in a gluster volume, and the app says it cannot -- it calculated it requires 15quintillion bytes. Sounds like gluster is reporting a wrong block size in a system call somewhere
16:37 vimal joined #gluster
16:41 justinmburrous joined #gluster
16:59 justinmb_ joined #gluster
17:04 theron_ joined #gluster
17:11 rwheeler joined #gluster
17:22 cjanbanan joined #gluster
17:30 DV joined #gluster
17:31 bennyturns joined #gluster
17:32 justinmburrous joined #gluster
17:38 cjanbanan joined #gluster
17:43 ekuric joined #gluster
17:44 dtrainor joined #gluster
17:48 ricky-ticky joined #gluster
18:05 DV joined #gluster
18:09 cjanbanan joined #gluster
18:09 giannello joined #gluster
18:12 justinmburrous joined #gluster
18:31 cjanbanan joined #gluster
18:31 justinmburrous joined #gluster
18:36 nueces joined #gluster
18:41 calum_ joined #gluster
18:46 julim joined #gluster
18:53 gildub joined #gluster
18:55 stickyboy joined #gluster
18:58 _dist joined #gluster
19:06 giannello joined #gluster
19:17 xleo joined #gluster
19:24 theron joined #gluster
19:26 nshaikh joined #gluster
19:28 cjanbanan joined #gluster
19:29 ryao joined #gluster
19:30 ryao Who can edit this? http://www.gluster.org/community/documentation/index.php/GlusterOnZFS#Finish_ZFS_Configuration
19:30 ryao Actually, I will go to #gluster-dev.
19:31 davemc ryao, do you have specific edits in mind?
19:31 lpabon joined #gluster
19:36 justinmburrous joined #gluster
19:38 fubada joined #gluster
19:39 fubada purpleidea: hi. im trying to get your vagrant up, however im running into this issue http://fpaste.org/140394/12797188/
19:39 glusterbot Title: #140394 Fedora Project Pastebin (at fpaste.org)
19:39 fubada can you advise?
19:47 ryao davemc: There are several things that I could improve there, but the main one that should be done ASAP is deleting the suggestion to do zfs_nocacheflush=1. That is incredibly dangerous.
19:49 davemc ryao, so if I delete just that param, we can continue to review it?
19:59 ryao davemc: I am not sure if I understand the question, but I am more than happy to help fix this. I just happened to hear about it at a time when I am somewhat strained for time. I am trying to get a large amount of things done before I had to LinuxCon Europe 2014.
20:00 davemc ryao, Until we can really fix the page, I'm trying to figure out the minimum I can do to avoid user data loss.
20:01 davemc so I was aksing if removing this "Ignore client flush/sync commands; let ZFS handle this with the transaction group timeout flush. NOTE: Requires a UPS backup solution unless you don't mind losing that 5 seconds worth of data. "
20:01 ryao davemc: Delete that parameter and do not suggest setting sync=disabled. Avoiding those settings will ensure that ZFS enforces proper POSIX semantics.
20:01 davemc and the actual "zfs_nocacheflush=1" suffices
20:01 davemc k
20:02 ryao That explanation is somewhat confused. zfs_nocacheflush=1 will ignore flushes between ZFS and the block layer. sync=disabled will ignore O_SYNC/fsync()/fdatasync() at the VFS layer.
20:02 bene2 joined #gluster
20:03 ryao Ensuring data is safe requires not doing either of them. The last 5 seconds bit is correct for sync=disabled though.
20:05 ryao The idea that the default zfs_txg_timeout is 30 is likely wrong. I would need to check the commit history to be certain, but the default was 5 as far back as I can remember.
20:06 davemc ryao, removed the sync=disabled bit, removed the nocacheflush param
20:06 ryao davemc: Thanks.
20:07 davemc if you see anything else, yell my way.
20:08 ryao davemc: There are other mistakes, but those can wait for another time. They don't jeoparodize data integrity.
20:08 davemc no problem, thanks for catching this.
20:09 ryao davemc: Thanks for caring. It is always nice to meet someone else who values data integrity. :)
20:09 _dist ryao: I can confirm the default txg is 5 now
20:09 davemc been on the wrong side of that before
20:10 semiosis that's what replication is for :)
20:11 _dist davemc: as someone who runs gluster on ZFS I would highly recommend using xattr=sa on datasets, without it ZFS will create a separate file for each xattr which could really slow things down in some cases
20:11 calisto1 joined #gluster
20:12 davemc _dist there seems to be a warning on xattr=sa under the Build and Install
20:13 _dist is it related to symlinks? there was a bug at one point with xattrs on symlinks I believe that was corrected
20:14 davemc _dist, don't know.  Can add it to an action to look into
20:15 _dist davemc: sounds good, the performance difference is significant as I'm sure you can imagine, especially in parity raid where files and their xattrs may be on different physical disks with xattr=on
20:16 davemc I bet it makes quite a difference at that
20:16 jansegre joined #gluster
20:17 giannello joined #gluster
20:19 nueces joined #gluster
20:19 plarsen joined #gluster
20:20 cjanbanan joined #gluster
20:27 sprachgenerator joined #gluster
20:28 dtrainor joined #gluster
20:29 sprachgenerator joined #gluster
20:30 dtrainor joined #gluster
20:39 purpleidea fubada:
20:39 purpleidea The paste you are looking for does not exist
20:45 julim joined #gluster
20:53 n-st joined #gluster
20:54 MacWinner joined #gluster
20:56 semiosis purpleidea: i saw the paste earlier, it was a syntax error about something like unescaped < in attribute somthing="#<Object...
20:56 semiosis in a network config iirc
20:57 purpleidea semiosis: oh i think that person emailed me then. i replied. thanks!
20:57 semiosis yw
20:57 purpleidea semiosis: btw, re: my gluster-devel reply to jdarcy... any idea about the status of ssl in gluster, or if you've used it?
20:58 semiosis haven't seen the post, but no idea about ssl
20:58 MacWinner joined #gluster
20:58 semiosis never looked into it
20:58 purpleidea cool thanks
20:58 semiosis yw
21:02 davemc joined #gluster
21:06 Telsin davemc: as another zfs under gluster user, the xattr=sa was for 0.6.2 and older builds of zfsol and have been corrected in 0.6.3,
21:06 theron joined #gluster
21:07 Telsin bloody thing bit me and I rebuild whole pools to get over it, too. arg.
21:07 Telsin definitely worth advising 0.6.3 or latest zfsol  in general
21:14 semiosis Telsin: feel free to add that to the wiki
21:15 semiosis that page (http://www.gluster.org/community/documentation/index.php/GlusterOnZFS) is by & for people who use gluster on zfs
21:18 Telsin I'll do that
21:18 semiosis thanks!
21:18 Telsin you guys know about this: "WARNING: Asirra, the cat and dog CAPTCHA, is closing permanently on October 6, 2014. "?
21:18 Telsin when creating an account
21:19 semiosis Telsin: yes the wiki admin is aware, thanks
21:19 semiosis JustinClift: ^^
21:19 Telsin figured you did, but… still works, fyi
21:20 cjanbanan joined #gluster
21:31 quique joined #gluster
21:36 justinmburrous joined #gluster
21:41 julim joined #gluster
21:44 theron_ joined #gluster
22:06 siel joined #gluster
22:09 justinmburrous joined #gluster
22:13 cjanbanan joined #gluster
22:27 jansegre joined #gluster
22:31 cjanbanan joined #gluster
22:51 siel joined #gluster
22:58 cjanbanan joined #gluster
23:00 harish joined #gluster
23:05 firemanxbr joined #gluster
23:17 cjanbanan joined #gluster
23:29 cjanbanan joined #gluster
23:34 siel joined #gluster
23:48 siel joined #gluster
23:59 plarsen joined #gluster

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