Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster, 2017-05-31

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

All times shown according to UTC.

Time Nick Message
00:17 masber joined #gluster
00:26 masber joined #gluster
00:27 Jacob843 joined #gluster
00:28 pioto_ joined #gluster
00:42 Alghost joined #gluster
01:07 vbellur joined #gluster
01:25 shdeng joined #gluster
01:30 cholcombe JoeJulian: interesting thread about gluster being removed from cinder.  I thought redhat knew about that and it was intentional
01:33 riyas joined #gluster
01:48 susant joined #gluster
01:50 farhorizon joined #gluster
01:53 kpease joined #gluster
02:02 derjohn_mob joined #gluster
02:06 amye cholcombe, the communication didn't make it through. hence the thread.
02:12 vbellur JoeJulian, amye: maybe we should determine how many users/deployments are going to be affected by unavailability of the cinder driver. That could help in determining our future course of action.
02:12 amye vbellur, the tricky part is that we don't measure anyone's deployments. :D
02:12 vbellur amye: we have a real opportunity to do that now. :D
02:13 amye Erf. Slippery slope.
02:13 amye Would be better off working with RDO and Openstack community
02:13 amye Start there, expand out to gluster.org overall as needed.
02:14 vbellur amye: what's the best way to engage in those communities? MLs?
02:14 amye vbellur: I'll ask Rich, RDO lead where that starts from.
02:15 amye vbellur, I'll let him know why we need halp :)
02:15 vbellur amye: that would be a good start :)
02:16 amye I mean, if anyone else has other ideas of what else we should be looking at, I'm all ears.
02:56 kramdoss_ joined #gluster
03:05 farhorizon joined #gluster
03:10 ppai joined #gluster
03:29 Humble joined #gluster
03:30 Shu6h3ndu joined #gluster
03:31 nbalacha joined #gluster
03:36 apandey joined #gluster
03:37 riyas joined #gluster
03:40 susant joined #gluster
03:43 hgowtham joined #gluster
03:50 skumar joined #gluster
03:51 atinm joined #gluster
03:52 om2 joined #gluster
03:56 susant left #gluster
04:10 jiffin joined #gluster
04:13 sanoj joined #gluster
04:15 susant joined #gluster
04:19 Karan joined #gluster
04:21 zbb9231 joined #gluster
04:26 DV joined #gluster
04:29 rwheeler joined #gluster
04:35 buvanesh_kumar joined #gluster
04:50 fcoelho joined #gluster
05:05 zbb9231 Hi all, been working on a small cluster and wondering if I have gluster set to replica 2 and add 2 more bricks does that mean the next time I want to extend I have to add 4 bricks?
05:09 jiffin joined #gluster
05:10 itisravi joined #gluster
05:10 Prasad joined #gluster
05:11 rafi joined #gluster
05:13 kdhananjay joined #gluster
05:14 susant zbb9231: if replica is set to 2, you will have to add bricks in pairs (2s) everytime
05:15 karthik_us joined #gluster
05:25 ndarshan joined #gluster
05:25 ashiq joined #gluster
05:27 rafi1 joined #gluster
05:31 skoduri joined #gluster
05:56 _KaszpiR_ joined #gluster
05:56 rastar joined #gluster
05:59 aravindavk joined #gluster
06:08 sona joined #gluster
06:13 poornima_ joined #gluster
06:18 om2 joined #gluster
06:30 jtux joined #gluster
06:32 jbrooks joined #gluster
06:34 Saravanakmr joined #gluster
06:40 mbukatov joined #gluster
07:03 kotreshhr joined #gluster
07:08 Humble joined #gluster
07:16 ivan_rossi joined #gluster
07:25 saintpablo joined #gluster
07:26 saintpablos joined #gluster
07:37 jkroon joined #gluster
07:43 fsimonce joined #gluster
07:45 renout_away joined #gluster
07:45 Alghost_ joined #gluster
08:02 Karan joined #gluster
08:24 miller92 joined #gluster
08:25 miller92 Hi all. Hope this is the right place to ask this question. I've been using gluster somewhat successfully but have a question regarding using ip's to setup bricks rather than hostnames
08:26 miller92 Currently to create a brick on two machines, they both have to have each other listed in their /etc/hosts file
08:26 miller92 so for example, "10.100.100.10 server"
08:26 miller92 and "10.100.100.11 server2"
08:26 miller92 ideally I want to just use the IP instead of the hostname, but I get pre-validation errors
08:33 miller92 it also seems that to add a new brick on a new server, I have to update all server's /etc/hosts file with the new server details
08:51 ajph_ joined #gluster
08:53 p7mo_ joined #gluster
08:53 d4n13L_ joined #gluster
08:55 kkeithle joined #gluster
08:56 lalatend1M joined #gluster
08:58 rastar miller92: IPs should work just fine. you don't need hostnames
08:59 mlg9000_1 joined #gluster
08:59 rastar miller92: what was the prevalidation error you encountered?
08:59 Chinorro joined #gluster
09:00 MadPsy_ joined #gluster
09:00 MadPsy_ joined #gluster
09:00 portante_ joined #gluster
09:00 marlinc_ joined #gluster
09:04 moneylotion joined #gluster
09:04 DaKnOb joined #gluster
09:07 Wizek_ joined #gluster
09:10 miller92 Ah apologies, it's a different error. Using 'gluster volume create volume1 replica 2 transport tcp 10.136.14.85:/gluster-storage 10.136.7.166:/gluster-storage force'
09:11 miller92 I get 'volume create: volume1: failed: Host 10.136.14.85 is not in 'Peer in Cluster' state'
09:11 miller92 raster: I have ran 'peer probe 10.136.7.166' which was successful
09:16 Saravanakmr joined #gluster
09:19 MrAbaddon joined #gluster
09:33 ppai joined #gluster
09:41 ankitr joined #gluster
09:57 toshywoshy anybody have any suggestions to solve "bad frame in setup_rt_frame" or "Transport endpoint is not connected"
10:03 hybrid512 joined #gluster
10:18 Karan joined #gluster
10:20 Saravanakmr joined #gluster
10:32 shyam joined #gluster
10:34 kotreshhr joined #gluster
10:39 victori joined #gluster
10:43 pioto_ joined #gluster
10:54 susant joined #gluster
11:09 amosbird hello, should I use ceph or gluster for data storage? What are the main considerations?
11:20 Klas ceph is more complex, and is harder to get fully redundant, if even possible
11:20 Klas but has more advanced features
11:20 Klas it's diffci
11:21 Klas *difficult to pin down a clear winner in general, both of them have different strong and weak points, in everything
11:21 Klas so, a better question to ask is to describe your case and ask if glusterfs is a good fit, I would say
11:29 om2 joined #gluster
11:38 susant joined #gluster
11:38 susant left #gluster
11:38 Saravanakmr joined #gluster
11:39 victori joined #gluster
11:40 social joined #gluster
11:41 ahino joined #gluster
11:49 om2 joined #gluster
11:51 * kkeithle wonders what "more advanced features" Ceph has that GlusterFS doesn't also have.
11:52 baber joined #gluster
11:57 ahino joined #gluster
12:02 jiffin1 joined #gluster
12:08 ppai joined #gluster
12:11 jkroon joined #gluster
12:15 rafi1 joined #gluster
12:20 Klas kkeithle: tiered storage is the only thing at the top of my head
12:21 Klas but, you can of course get such features behind glusterfs on the OS or baremetal level
12:21 Klas direct block storage is also possible, unless I'm mistaken
12:22 Klas but I've only evaluated ceph quite lightly, it always seemed to have to many parts which could be broken
12:22 kkeithle glusterfs has tiering, as of 3.8.
12:23 Klas oh
12:23 kkeithle gluster-block is a (new) things
12:23 kkeithle thing
12:23 Klas didn't know that =)
12:24 kkeithle tiering as of 3.7 actually
12:24 aravindavk joined #gluster
12:31 vbellur joined #gluster
12:31 pioto_ joined #gluster
12:36 Klas kkeithley: huh, then I missed it when we implemented glusterfs
12:38 _KaszpiR_ [13:20:18] <Klas> ceph is more complex, and is harder to get fully redundant, if even possible
12:38 _KaszpiR_ really?
12:38 _KaszpiR_ thats interesting
12:38 Alghost joined #gluster
12:38 _KaszpiR_ could you elaborate more?
12:38 Klas not really
12:38 Klas I remember having that impression
12:38 Klas but I've never implemented ceph
12:38 _KaszpiR_ ah ok
12:38 Klas we split teams at work
12:38 Klas I wanted glusterfs
12:39 kkeithley I suspect a lot of people agree that Ceph is more complex to set up
12:39 Klas and the results seemed nicer from my team than the ceph team
12:39 Klas basically
12:39 _KaszpiR_ uhum
12:39 _KaszpiR_ but how big sotrage can you stuff into gluster?
12:39 kkeithley and for a lot of workloads Gluster performs better
12:39 ajph joined #gluster
12:40 jiffin1 joined #gluster
12:40 _KaszpiR_ if it is easier, then why it was removed from openstack?
12:41 ajph joined #gluster
12:41 shyam joined #gluster
12:41 kkeithley someone said the cinder driver wasn't being maintained. I don't know the real reason
12:42 _KaszpiR_ uhum
12:42 kkeithley You could easily set up 64 2U servers with 4x8TB drives. You can do the math. I've heard of petabyte deployments of gluster
12:42 _KaszpiR_ but then is gluster rack aware?
12:42 Klas and files can be split between servers
12:42 kkeithley mangement starts to get hard at that scale
12:42 _KaszpiR_ yeah no wonder
12:43 kkeithley Fixing the management is a GlusterFS 4.0 feature
12:43 Klas _KaszpiR_: how do you mean rack aware?
12:43 _KaszpiR_ say I have 4 racks and I want to be sure that data is not replicated across the same rack but in different racks
12:44 kkeithley dunno, is Ceph rack aware?  I kinda suspect it isn't, but I don't know that for certain.
12:44 _KaszpiR_ I'd have to do it with brick ordering, AFAIR
12:44 _KaszpiR_ not sure about ceph
12:44 _KaszpiR_ but for example with elasticsearch it is possible
12:44 Klas no idea about scales in that order
12:45 _KaszpiR_ what I want to say is that it would be nice to have a specific parameter which would be picked up by gluster when adding brick
12:45 _KaszpiR_ so that it would not be striclty tied to brick order when adding them
12:46 Klas yeah, it does sound cool
12:46 _KaszpiR_ I guess this could be done outsied of the gluster itself
12:47 _KaszpiR_ for example using facter facts create some script logic which whould say detect specific parameter and reorder randomly added servers
12:47 nbalacha joined #gluster
12:47 _KaszpiR_ let say take rack fackt as input, and let say 8 servers in not particular order, but each would have specific fact set about rack it is placed in
12:48 _KaszpiR_ then the script would order the list and feed it to gluster
12:48 _KaszpiR_ hm should be possible do to it with some puppetdb api or something
12:48 * kkeithley tries to remember if heketi does rack aware volume config
12:49 miller92 left #gluster
12:50 _KaszpiR_ from other concers, I'm surprised gluster does not allow joining servers based on somekind of secret
12:50 _KaszpiR_ right now adding seerver must be triggered from one of the nodes withing cluster
12:50 _KaszpiR_ It would be much more convenient if given new server could join pool if it knew secret used by that cluster
12:51 _KaszpiR_ are there any onsiderations about implementing this in the future?
12:52 Klas contributions are always appreciated ;)?
12:52 ankitr joined #gluster
12:52 _KaszpiR_ just like spare time? :D
12:53 Klas as I said, I have never considered that scale, we are only using it for a few TB across a couple of halls
12:53 rafi joined #gluster
12:54 _KaszpiR_ yeah, same here, can you share what you use for config management?
12:54 Klas but I always like the idea of making people consider creating great functions yourself
12:54 Klas our orchestration is about to be phased out, it's a weird hodgepodge of various perl and bash scripts
12:54 Klas some if it stretches back before orchestration was really a thing
12:55 Klas it works very well, but it would take several hours explaining it
12:55 Klas and it's not perfect
12:55 Klas currently, with gluster, we do most things manually
12:58 Klas (I work for a university, and I have about 20 entirely different things to do)
12:58 tru_tru joined #gluster
12:59 _KaszpiR_ interesting
12:59 _KaszpiR_ how about ansible/puppet?
12:59 Klas we are looking at salt
12:59 _KaszpiR_ ah, good choice
13:00 Klas yeah, it does seem nice
13:00 Klas seems to be able to do everything salt+puppet is capable of
13:00 Klas and seems cleaner
13:00 _KaszpiR_ 'salt+puppet' or you mean ansible+puppet?
13:00 buvanesh_kumar joined #gluster
13:00 Klas haha
13:01 Klas I meant that, right ;)
13:01 Klas just tired
13:01 Klas oh, look at that, I can go home
13:01 Klas cya ;)
13:01 _KaszpiR_ thankx for info, bye :D
13:18 kotreshhr left #gluster
13:22 skylar joined #gluster
13:27 dijuremo What changes may have affected my performance by going from 3.6.6 to 3.10.2 ? I was doing nightly rsync to backup the data from the gluster volume and now this takes 2-3 hours extra.
13:28 dijuremo Are there any volume options I should be adjusting to get back the lost performance?
13:30 dijuremo https://thepasteb.in/p/KOh8pB5xjoktJ -> My current volume options
13:30 glusterbot Title: ThePasteBin - For all your pasting needs! (at thepasteb.in)
13:31 jstrunk joined #gluster
13:33 jkroon joined #gluster
13:51 vbellur joined #gluster
13:55 ankitr joined #gluster
13:56 Jacob843 joined #gluster
14:06 skylar joined #gluster
14:19 mlhess joined #gluster
14:22 shyam joined #gluster
14:25 farhorizon joined #gluster
14:35 ic0n joined #gluster
14:36 nbalacha joined #gluster
14:46 dijuremo Anyone can shime in on my upgrade issue from 3.6.6 to 3.10.2?
14:51 dijuremo I also asked a questions about another issue with Revit .rvt files that seems affected after upgrade from 3.6.6 to 3.10.2 which has gone unanswered, is there anyone willing to try and help?
14:53 dijuremo After our upgrade on the server from 3.6.6 to 3.10.2, users accessing .rvt files via samba with gluster vfs plugin, get an ACCESS DENIED error from the application itself. We have found a workaround, but would like to try an understand what is happening.
14:54 dijuremo Our work around is: Rename file MyFile.rvt to MyFile2.rvt using windows explorer, then rename it back, then it opens OK. Not sure if the rename operation is forcing some attribute upgrade, but I really do not know about gluster to look into this much. I can provide logs, etc if anyone can point me in the direction I should be looking for answers.
15:02 _KaszpiR_ dijuremodid you try sending a message to mailing list?
15:03 kpease joined #gluster
15:03 _KaszpiR_ I think it will get better attention there than on IRC
15:05 wushudoin joined #gluster
15:06 shyam joined #gluster
15:12 rastar dijuremo: I can't help you with performance related issue but I can with gluster+samba
15:12 rastar dijuremo: which version of samba are you using?
15:13 Shu6h3ndu joined #gluster
15:15 susant joined #gluster
15:19 dijuremo _KaszpiR_ Yes I did sen to the mailing list and nada...
15:19 dijuremo The file issue yesterday and the other issue about rsync speed I sent today.
15:20 dijuremo http://lists.gluster.org/pipermail/gluster-users/2017-May/031293.html
15:20 glusterbot Title: [Gluster-users] URGENT: Update issues from 3.6.6 to 3.10.2 Accessing files via samba come up with permission denied (at lists.gluster.org)
15:21 dijuremo It is strange that the problem is only related to files used by Autodesk's Revit program. Both .rvt (project files) and .rfa (family files) seem affected, but it is weird that some files work just fine and some give the ACCESS DENIED error...
15:26 _KaszpiR_ maybe they are locked/used by users?
15:26 _KaszpiR_ or maybe somekind of shadow copies/snapshots?
15:28 buvanesh_kumar joined #gluster
15:30 dijuremo No snapshots, not locked by users.. The user can rename them via explorer, only when they try to open them from Revit they get the ACESS DENIED error.
15:30 dijuremo Also on the server itself, as the user trying to access the file, I can do a strings RevitFile.rvt and it comes through the screen....
15:31 dijuremo What is the best way to attach logs requested on the mailing list? I have a tar.bz2 file that is 1.6MB Not sure I should send it to the whole list....
15:35 dijuremo I guess a google drive link is best than spamming the whole list... did just that... :)
15:36 om2 joined #gluster
16:07 rastar dijuremo: I replied in email.
16:15 linuxaddicts joined #gluster
16:20 [o__o] joined #gluster
16:29 Jacob843 joined #gluster
16:32 Gambit15 joined #gluster
16:32 JoeJulian Good local greeting time everybody.
16:35 farhorizon joined #gluster
16:46 bwerthmann joined #gluster
16:47 [o__o] joined #gluster
16:48 jiffin joined #gluster
16:51 apandey joined #gluster
16:53 farhorizon joined #gluster
16:53 apandey_ joined #gluster
16:54 ivan_rossi left #gluster
16:55 riyas joined #gluster
17:02 Saravanakmr joined #gluster
17:07 victori joined #gluster
17:08 apandey__ joined #gluster
17:08 Karan joined #gluster
17:12 Wizek_ joined #gluster
17:14 kpease joined #gluster
17:21 semiosis_ joined #gluster
17:34 Jacob843 joined #gluster
17:49 Jacob843 joined #gluster
17:54 vbellur joined #gluster
17:56 Jacob8432 joined #gluster
18:16 ahino joined #gluster
18:18 MadPsy joined #gluster
18:18 MadPsy joined #gluster
18:20 Ashutto joined #gluster
18:24 baber joined #gluster
18:28 farhorizon joined #gluster
18:45 kovshenin joined #gluster
18:56 farhorizon joined #gluster
18:58 skoduri joined #gluster
19:14 buvanesh_kumar joined #gluster
19:35 farhorizon joined #gluster
19:42 baber joined #gluster
19:57 jkroon joined #gluster
20:04 bwerthmann joined #gluster
20:13 cornfed78 joined #gluster
20:17 baber joined #gluster
20:20 sadbox joined #gluster
20:41 bartden joined #gluster
20:42 bartden hi, can i mount a gluster server 3.7.5 with gluster 3.10 client?
20:44 farhorizon joined #gluster
21:26 JoeJulian bartden: Generally you want to upgrade servers before clients. I don't know of any specific problems between those versions though.
21:30 shyam joined #gluster
21:31 shyam left #gluster
21:47 jarbod_ joined #gluster
21:50 vbellur joined #gluster
21:52 vbellur joined #gluster
21:52 panina joined #gluster
21:57 cherrysuckle joined #gluster
22:07 farhorizon joined #gluster
22:12 cherrysuckle joined #gluster
22:16 devyani7 joined #gluster
22:24 vbellur joined #gluster
22:31 doc|work joined #gluster
22:31 doc|work hey, anyone using ubuntu/debian under docker/kubernetes had it that you get unknown filesystem glusterfs even though you have glusterfs-client installed? What did you do to fix it?
22:32 doc|work works as expected when I do it via 'mount'
22:38 wushudoin joined #gluster
23:17 doc|work joined #gluster
23:19 Alghost joined #gluster
23:24 ahino joined #gluster
23:29 MrAbaddon joined #gluster
23:33 doc|work joined #gluster
23:34 doc|work anyone?
23:46 masber joined #gluster
23:58 MrAbaddon joined #gluster

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