Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2014-10-15

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

All times shown according to UTC.

Time Nick Message
01:34 ws2k3 joined #gluster-dev
01:45 mikedep3- joined #gluster-dev
01:45 sac`away joined #gluster-dev
01:45 ws2k3 joined #gluster-dev
01:54 bala1 joined #gluster-dev
02:33 _Bryan_ joined #gluster-dev
03:50 shubhendu joined #gluster-dev
03:56 _ndevos joined #gluster-dev
03:56 itisravi joined #gluster-dev
04:05 soumya joined #gluster-dev
04:25 rafi1 joined #gluster-dev
04:26 Rafi_kc joined #gluster-dev
04:27 kdhananjay joined #gluster-dev
04:27 anoopcs joined #gluster-dev
04:34 jiffin joined #gluster-dev
04:39 atalur joined #gluster-dev
04:44 pranithk joined #gluster-dev
04:52 ppai joined #gluster-dev
04:53 kanagaraj joined #gluster-dev
05:08 atinmu joined #gluster-dev
05:15 ndarshan joined #gluster-dev
05:22 Gaurav__ joined #gluster-dev
05:24 lalatenduM joined #gluster-dev
05:28 nishanth joined #gluster-dev
05:30 spandit joined #gluster-dev
05:51 Humble joined #gluster-dev
06:00 shubhendu_ joined #gluster-dev
06:07 rgustafs joined #gluster-dev
06:19 shubhendu_ joined #gluster-dev
06:22 ndarshan joined #gluster-dev
06:23 raghu joined #gluster-dev
06:35 RaSTar joined #gluster-dev
06:42 kshlm joined #gluster-dev
06:59 shubhendu_ joined #gluster-dev
07:01 ndarshan joined #gluster-dev
07:02 atalur joined #gluster-dev
07:09 atalur joined #gluster-dev
07:14 atalur joined #gluster-dev
07:19 atinmu joined #gluster-dev
07:20 pranithk joined #gluster-dev
07:28 rgustafs joined #gluster-dev
08:05 pranithk joined #gluster-dev
08:12 atinmu joined #gluster-dev
08:30 aravindavk joined #gluster-dev
08:50 nishanth joined #gluster-dev
08:51 ndarshan joined #gluster-dev
08:57 vikumar joined #gluster-dev
09:05 itisravi joined #gluster-dev
09:11 deepakcs joined #gluster-dev
09:42 spandit joined #gluster-dev
09:43 lalatenduM joined #gluster-dev
09:52 ndarshan joined #gluster-dev
09:52 nishanth joined #gluster-dev
09:53 RaSTar joined #gluster-dev
10:10 lalatenduM_ joined #gluster-dev
10:14 RaSTar joined #gluster-dev
10:21 atinmu joined #gluster-dev
10:22 Gaurav__ joined #gluster-dev
10:45 spandit joined #gluster-dev
10:50 pranithk left #gluster-dev
10:56 deepakcs joined #gluster-dev
11:00 edward1 joined #gluster-dev
11:00 ppai joined #gluster-dev
11:01 ppai joined #gluster-dev
11:23 ira joined #gluster-dev
11:37 Gaurav__ joined #gluster-dev
11:37 atinmu joined #gluster-dev
11:44 _ndevos joined #gluster-dev
11:56 tdasilva joined #gluster-dev
12:06 pranithk joined #gluster-dev
12:07 kdhananjay joined #gluster-dev
12:08 kanagaraj joined #gluster-dev
12:12 pranithk kdhananjay: Did you get a chance to review http://review.gluster.com/#/c/8918/
12:13 kdhananjay pranithk: I took a look at the patch. Need to go through the bug description once to know what it is trying to solve.
12:13 pranithk kdhananjay: cool!
12:13 rgustafs joined #gluster-dev
12:15 pranithk ndevos: I need to talk to you after the gluster-meeting
12:16 ndevos pranithk: ok?
12:16 pranithk ndevos: It is about pacemaker. For nfs ha
12:17 ndevos pranithk: okay, talk to me, or to kkeithley?
12:17 pranithk ndevos: we can chat here, kkeithley can also comment I guess. I have some doubts which I wanted clarified. Nothing major
12:18 ndevos pranithk: sure
12:47 ira joined #gluster-dev
12:58 tdasilva joined #gluster-dev
13:04 lalatenduM joined #gluster-dev
13:06 RaSTar joined #gluster-dev
13:15 pranithk ndevos: kkeithley: I heard you guys are using pacemaker for ha purposes in nfs-ganesh + gluster. Could you let me know more about it? Is there any document?
13:16 ndevos pranithk: that really is mostly kkeithley's doing :)
13:16 kkeithley yeah, one sec
13:16 pranithk kkeithley: I would like to know more about it (I know nothing about it :-D ). What kind of ha does it give?
13:16 * ndevos leaves, might be back later
13:17 kkeithley we're doing active-active, but it'll do active-passive as well.
13:17 kkeithley here's a good intro: http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/pdf/Clusters_from_Scratch/Pacemaker-1.1-Clusters_from_Scratch-en-US.pdf
13:17 pranithk kkeithley: Okay the main doubt I have is, how does it achieve ha? where does the client reside and where does nfs-ganesha server reside?
13:18 ppai joined #gluster-dev
13:19 pranithk kkeithley: java?
13:19 kkeithley basically on a four server cluster we run ganesha  on all four nodes. We use a virtIP on each node, and if ganesha goes down on a node we fail the virtIP to another node. (nfs clients connect to the virtIP)
13:19 kkeithley no java involved
13:19 kkeithley AFAIK
13:20 kkeithley pacemaker and corosync
13:20 Humble treat virtual ip == floating ip
13:20 Humble pranithk, ^^^
13:20 kkeithley right
13:21 kkeithley Anand's design. I can forward his notes to you.
13:21 Humble kkeithley, can u Cc me in
13:21 pranithk kkeithley: oh that would be good.
13:21 kkeithley yup
13:21 Humble thanks. kkeithley++
13:21 glusterbot Humble: kkeithley's karma is now 24
13:21 pranithk kkeithley: I want to know what will happen for O_APPEND writes/entry-operations when the same operation is retried
13:22 kkeithley that's a good question
13:22 pranithk kkeithley: Any chance you may know?
13:23 kkeithley I don't know. That's what makes it a good question. ;-_)
13:23 kkeithley ;-)
13:23 pranithk kkeithley: nice :-)
13:23 kkeithley but part of the answer is...
13:24 kkeithley when we fail-over the virtIP, we put all the ganesha servers into grace
13:24 kkeithley I'm not fully up on NFSv4 protocol, but I'd guess that might inhibit retries
13:25 kkeithley Anand probably knows the answer
13:25 pranithk kkeithley: cool
13:25 pranithk kkeithley: do you guys have some alias I can send a mail to?
13:25 kkeithley during grace all the locks are supposed to get resolved, among other things
13:25 kkeithley nfs-ganesha@redhat.com
13:26 pranithk kkeithley: cool
13:45 pranithk Humble: You have experience about pacemaker?
13:46 kshlm joined #gluster-dev
13:47 shyam joined #gluster-dev
13:50 pranithk kkeithley: It doesn't seem to have details :-(
13:50 pranithk kkeithley: I sent a mail to the mailer list. Let me see.
13:50 pranithk kkeithley: Wow I got a reject from the owner of the mailing list
13:51 kkeithley yes, it's pretty sparse on details
13:51 kkeithley probably need to subscribe first
13:51 pranithk kkeithley: I will send it to individuals. Anands, you, ndevos, soumya, meghna, jiffin. Anyone else?
13:51 kkeithley that's it
13:52 pranithk kkeithley: cool
13:52 dlambrig_ joined #gluster-dev
13:57 pranithk ndevos: Wow, you are at linuxcon?
14:23 jbautista- joined #gluster-dev
14:29 _Bryan_ joined #gluster-dev
14:37 davemc joined #gluster-dev
15:15 lalatenduM joined #gluster-dev
16:06 jobewan joined #gluster-dev
16:08 pranithk soumya: Could you explain me basics of nfs-ganesha tomorrow once?
16:12 semiosis kkeithley: pranithk: did you see in #gluster that Slydder is testing out ganesha today?  not doing HA, just interested in performance & localhost mount without deadlocking
16:12 semiosis sounds like ganesha uses libgfapi, not a fuse mount
16:13 semiosis https://botbot.me/freenode/gluster/search/?q=ganesha
16:29 ndevos pranithk: yes, I am, and now getting ready for a dinner :)
16:29 kkeithley correct, ganesha's FSAL_GLUSTER uses gfapi
16:38 soumya pranithk, sure :)
16:43 soumya semiosis, its really great to hear that Slydder has seen performance improvement while using ganesha. It would be interesting to know the test details and the results you got.
16:43 kkeithley And we just tested ganesha at NFS bake-a-thon last week. Latest ganesha (2.2dev14) and reasonable current $HEAD of master branch. Quite successfully too, I might add
16:44 soumya Could you or Slydder send us the test details and the results , maybe to gluster-devel mailing list itself.
16:44 ndevos kkeithley: re: rhel-6.6 issue, I think we should provide a gluster-release package (one for each branch, 3.4, 3.5, 3.6, master) that contains a .repo file and depends on yum-plugin-priorities
16:45 semiosis i'm just reporting what i've heard in #gluster, i'm not involved in Slydder's tests
16:45 semiosis soumya: i'll ask tho
16:45 soumya semiosis, ok thanks
16:45 semiosis seems he's left the channel.  i'll ask later
16:45 soumya yeah
16:46 ndevos soumya: he seems to be active during emea hours, I think he's in germany - you can probably ping him yourself tomorrow
16:46 kkeithley ndevos: yes, that would be good
16:46 soumya ndevos, yeah sure
16:48 semiosis is ganesha the long term plan for nfs in gluster?
16:48 soumya kkeithley, ndevos, could you please review my README & gluster config file changes
16:49 kkeithley soumya: sure
16:49 ndevos soumya: I'll try to have a look at that tomorroe
16:49 ndevos *tomorrow
16:49 semiosis i mean, will gluster-nfs become deprecated in a future release in favor of ganesha?
16:49 ndevos semiosis: yeah, something like that
16:49 soumya kkeithley, ndevos, thanks..I want to get that patch pushed this week as there are users getting mislead with older deprecated options
16:49 kkeithley semiosis: it is for RHS. I suppose for community gluster too, although I don't believe there's been any reasl discussion along those lines
16:50 ndevos kkeithley: I hope to deprecate gNFS too, but have not thought about a plan or timeline yet
16:50 semiosis interesting
16:51 semiosis i'm excited about the possibility of pnfs clients with gluster servers
16:51 ndevos there will be very few feature enhancements in gluster/nfs, most work will be put in ganesha (it has NFSv4 and pnfs and all already, re-implementing that in gluster/nfs does not make much sense)
16:51 semiosis seems like that would solve the NFS HA problem
16:51 kkeithley RHS will eventually deprecate the gnfs server xlator and Red Hat will probably stop allocating dev resources to it.
16:51 * semiosis not a fan of floating IPs
16:52 ndevos semiosis: jiffin mentioned today that he is looking into that, I encouraged him to send a notice to the mailinglists to let others know about the progress
16:52 semiosis sweet
16:52 kkeithley Implementing NFSv4 in gluster would be a big job. Why reinvent the wheel?
16:52 semiosis +1
16:54 kkeithley Well, until ganesha has pNFS... The first, or interim, HA solution will use floating IPs, managed by pacemaker+corosync
16:54 kkeithley just to set expections
16:55 kkeithley expectations
16:55 semiosis right
16:56 kkeithley It's all a bit tentative right now though; things could change
16:56 ndevos kkeithley: btw, https://forge.gluster.org/bugzappers/nightly-builds/blobs/master/dgo-nightly.spec is a gluster-release package for the nightly builds, I guess we can base a 'stable' version on that
16:57 ndevos kkeithley: I'll leave that hangin' like this for now, maybe you want to follow up on that, or I could try over the next few days
16:57 * ndevos goes for dinner, ttyl!
16:57 kkeithley okay, I can cobble something up in a bit. Let me finish what I'm working on, then I'll switch gears
16:59 Gaurav_ joined #gluster-dev
17:29 davemc joined #gluster-dev
17:29 charta joined #gluster-dev
17:34 kkeithley soumya: README and gluster.conf look good.
17:35 soumya kkeithley, thanks. Frank has given a comment saying that we should take out defaults (which are commented in the README file). But I am not 100% convinced whether we should do so though he has a valid point that we should keep it updated everytime a option gets removed/modified.
17:36 soumya kkeithley, what do you suggest?
17:42 kkeithley bleh. It's barely different than vfs.conf.  We have Squash and SecType (compared to vfs.conf). I think it "should just work" out of the box, except for changing the paths and gluster volname. I think it's fine as is.
17:52 soumya right thats about gluster.conf. I added Squash and SecType which are essential to allow root permissions. I guess he was talking about the FSAL_GLUSTER/README file. But as you said, I too feel its fine as it is now.
17:53 soumya Will wait till tomorrow for any more comments and then send him Pull request. Thanks
17:56 systemonkey joined #gluster-dev
18:18 lpabon joined #gluster-dev
18:29 systemonkey joined #gluster-dev
18:38 kkeithley ndevos, nixpanic: see https://github.com/kalebskeithley/gluster-release
18:38 charta joined #gluster-dev
19:11 ilbot3 joined #gluster-dev
19:11 Topic for #gluster-dev is now Gluster Development Channel - http://gluster.org | For general chat go to #gluster | Patches - http://review.gluster.org/ | Channel Logs - https://botbot.me/freenode/gluster-dev/ & http://irclog.perlgeek.de/gluster-dev/
19:50 xavih joined #gluster-dev
20:16 charta joined #gluster-dev
21:10 davemc joined #gluster-dev
21:34 dlambrig_ left #gluster-dev
22:14 charta joined #gluster-dev
22:32 badone joined #gluster-dev

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