Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2015-11-09

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

All times shown according to UTC.

Time Nick Message
00:11 zhangjn joined #gluster-dev
00:56 zhangjn joined #gluster-dev
00:57 zhangjn joined #gluster-dev
01:02 zhangjn_ joined #gluster-dev
01:04 EinstCrazy joined #gluster-dev
01:04 EinstCrazy joined #gluster-dev
01:11 shaunm joined #gluster-dev
03:02 kdhananjay joined #gluster-dev
03:28 mohan_ joined #gluster-dev
03:29 sakshi joined #gluster-dev
03:32 Manikandan joined #gluster-dev
03:33 overclk_ joined #gluster-dev
03:44 itisravi joined #gluster-dev
03:45 itisravi joined #gluster-dev
03:48 itisravi joined #gluster-dev
03:51 gem joined #gluster-dev
03:53 atinm joined #gluster-dev
03:54 nbalacha joined #gluster-dev
03:55 bkunal joined #gluster-dev
03:58 anekkunt joined #gluster-dev
03:58 zhangjn joined #gluster-dev
04:00 zhangjn joined #gluster-dev
04:01 shaunm joined #gluster-dev
04:06 kdhananjay joined #gluster-dev
04:08 kdhananjay joined #gluster-dev
04:22 kshlm joined #gluster-dev
04:27 vimal joined #gluster-dev
04:32 kotreshhr joined #gluster-dev
04:32 kotreshhr left #gluster-dev
04:35 zhangjn joined #gluster-dev
04:41 Humble joined #gluster-dev
04:56 overclk joined #gluster-dev
05:02 kanagaraj joined #gluster-dev
05:06 hgowtham_ joined #gluster-dev
05:08 ndarshan joined #gluster-dev
05:09 pppp joined #gluster-dev
05:18 overclk_ joined #gluster-dev
05:19 rafi joined #gluster-dev
05:23 asengupt joined #gluster-dev
05:23 mchangir joined #gluster-dev
05:41 deepakcs joined #gluster-dev
05:43 skoduri joined #gluster-dev
05:55 Apeksha joined #gluster-dev
05:55 kotreshhr joined #gluster-dev
05:56 vmallika joined #gluster-dev
05:59 ashiq joined #gluster-dev
06:01 zhangjn_ joined #gluster-dev
06:04 aravindavk joined #gluster-dev
06:14 ashiq anoopcs++, thanks :)
06:14 glusterbot ashiq: anoopcs's karma is now 21
06:19 kotreshhr Mi4i has 5MP front cam
06:20 kotreshhr left #gluster-dev
06:20 kotreshhr joined #gluster-dev
06:20 nishanth joined #gluster-dev
06:21 spalai joined #gluster-dev
06:26 zhangjn joined #gluster-dev
06:27 hagarth joined #gluster-dev
06:29 shubhendu joined #gluster-dev
06:31 rraja joined #gluster-dev
06:35 zhangjn joined #gluster-dev
06:36 EinstCrazy joined #gluster-dev
06:38 ppai joined #gluster-dev
06:44 zhangjn joined #gluster-dev
06:49 atalur_ joined #gluster-dev
07:06 EinstCrazy joined #gluster-dev
07:18 zhangjn joined #gluster-dev
07:23 mmckeen joined #gluster-dev
07:26 zhangjn joined #gluster-dev
07:27 EinstCrazy joined #gluster-dev
07:30 mmckeen joined #gluster-dev
07:42 mmckeen joined #gluster-dev
07:53 zhangjn joined #gluster-dev
08:03 zhangjn joined #gluster-dev
08:03 overclk joined #gluster-dev
08:17 tigert Humble, hagarth: ping?
08:17 tigert whats the deal with the roadmap -> 3.8 doc?
08:18 tigert that link does not work, is that supposed to be a separate roadmap page or shouldnt it be part of the actual roadmap document?
08:18 tigert I can fix the link that I noticed was broken, but just wondering how that should be
08:19 kdhananjay1 joined #gluster-dev
08:22 tigert shouldnt we just combine those items into one page?
08:23 kdhananjay joined #gluster-dev
08:28 poornimag joined #gluster-dev
08:29 poornimag ashiq++ csim++ Humble++
08:29 glusterbot poornimag: ashiq's karma is now 14
08:29 glusterbot poornimag: csim's karma is now 15
08:29 glusterbot poornimag: Humble's karma is now 21
08:33 EinstCrazy joined #gluster-dev
08:33 zhangjn joined #gluster-dev
08:34 pranithk joined #gluster-dev
08:34 pranithk xavih: I am working on ec-readdir.t, I found one issue, but there still seems to be one more issue. I will let you know more about the bug as soon as I am done with it
08:35 hagarth tigert: pong, 3.8 can be found here should point to 3.8.md
08:37 zhangjn joined #gluster-dev
08:38 xavih pranithk: I've also found a possible cause, but I haven't had time to fully analyze it and fix it
08:39 tigert hagarth: yeah the link seems to be broken (for some reason maybe due to having two dots in the filename the resulted html file is called "3.8"
08:39 tigert so the link in https://www.gluster.org/community/roadmap/ is broken for me
08:40 tigert just pondering should I rename the file so the link works (you cannot link to .md because thats the source filename)
08:40 tigert it works in the github preview I guess but not in the website since those things get compiled to html
08:41 hagarth tigert: yes, I am trying to figure out how to make it work in both github and gluster.org
08:41 hagarth should we create a new directory called 3.8 and have an index.md in there?
08:41 hagarth with 3.8.renamed to 3.8/index.md ?
08:42 hagarth s/3.8.renamed/3.8.md renamed/
08:42 tigert I guess it is a problem with .md in general since the website is compiled, and github can display render the .md when you open the file in the git repo
08:43 hagarth tigert: right
08:43 tigert are you using the github side to something else than just as a website code repo?
08:44 hagarth tigert: no
08:49 pranithk xavih: sorry was on call. This is the issue I found. 1) 2+1 volume. 'd1' directory on Brick-0 is bad. 2) readdir takes locks and lock->good_mask is '7' 3) readdir does xattrop and fop->mask is '6' 4) ec_lock_update_good() will do lock->good_mask &= ~fop->mask | fop->remaining; which will set lock->good_mask to '1' 5) lock->good_mask |= fop->good will make it back to '7' so the reusing locks think all bricks are good...
08:50 pranithk xavih: wait, I messed up steps 4), 5) values. But I think you'll get the idea. Bad bricks is added back to lock->good_mask
08:50 bkunal Humble, hey do we have any guidelines for naming convention for doc?
08:52 overclk joined #gluster-dev
08:53 pranithk xavih: got it?
08:53 xavih pranithk: readdir does't update anything in ec_lock_update_good()
08:53 xavih pranithk: it has fop->expected == 1
08:56 pranithk xavih: Damn, you are right. But it is still the same issue. In the sense that the lock->good_mask is still '7' where as it should be '6'
08:56 xavih pranithk: yes. That's what I've found
08:56 pranithk xavih: I fixed that part and was testing it. And it failed again. But I don't remember if I compiled it or not. So doing it again
08:57 pranithk xavih: Did you find any other problem?
08:57 pranithk xavih: I was fixing it in ec_prepare_update_cbk()
08:57 pranithk xavih: Did you have any other fix in mind?
08:59 xavih pranithk: It seems that when ec_update_size_version() is called, lock->good_mask has not been correctly updates and sends the update even to the failed brick
09:00 xavih pranithk: In theory mkdir and touch should clear bits of good_mask
09:00 pranithk xavih: But that can happen only for fop->expected==1
09:01 pranithk xavih: otherwise it will work fine...
09:01 xavih pranithk: I've tried disabling client size self-heal and it seems the issue disappears, but I need to check if this is really the cause
09:02 hagarth tigert: what would be your recommendation? do we create a new folder - one per future release?
09:03 xavih pranithk: I'm thinking that maybe the ec_lock_update_good() corresponding to the heal fop could be clearing unwanted flags, specially for partial self-heals of directories
09:04 xavih pranithk: however I haven't had time to work on this yet
09:04 pranithk xavih: but self-heal doesn't use this framework at all right?
09:05 xavih pranithk: internally not, but it creates a fop
09:05 pranithk xavih: Sorry, not following you :-(.
09:07 Humble bkunal, not really ..
09:08 Humble as long as its in the right place and name conveys the content
09:08 bkunal Humble, ok
09:08 xavih pranithk: you are right, heal fop shouldn't update good_mask...
09:09 bkunal Humble, I am even moving other docs from docs/developer-guide to glusterdocs repo
09:09 xavih pranithk: I'll need to do more checks
09:09 pranithk xavih: I am not finding any  more issue other than that at the moment. Let me see. I compiled the code now and running it for a while now. It hadn't failed. Once that is done, I will send out the patch.
09:09 pranithk xavih: okay
09:11 Humble bkunal, the developer oriented information should stay in developer guide
09:11 bkunal Humble, ya sure
09:11 Humble only admin/user content should go inside glusterdocs admin repo
09:12 bkunal Humble, yes right, Had discussion with ndevos  on what to be moved and what not
09:12 pranithk xavih: I will run it for around an hour. It used to fail in 10 minutes max. Will that be fine?
09:12 pranithk xavih: If it still passes I will send the patch.
09:13 Humble bkunal++ , oh. ok..
09:13 glusterbot Humble: bkunal's karma is now 1
09:13 Humble ndevos++ :)
09:13 glusterbot Humble: ndevos's karma is now 213
09:13 bkunal Humble++
09:13 glusterbot bkunal: Humble's karma is now 22
09:13 xavih pranithk: Yes. Anyway I'll review the patch, so I'll check if it's the cause of the problem I've seen
09:13 bkunal ndevos++
09:13 glusterbot bkunal: ndevos's karma is now 214
09:55 Manikandan joined #gluster-dev
10:00 tigert hagarth: or keep one roadmap maybe?
10:00 tigert hagarth: currently there is 3.8 and nothing on 4.0 yet
10:01 tigert maybe if there is too much content to fit on one page, we could split it
10:06 pranithk xavih: There still seems to be one more issue. I see that it failed, but after 45 minutes instead of 10 minutes. Will look into that one as well...
10:07 hagarth tigert: I would prefer to have separate pages (primarily to have an uncluttered look)
10:10 tigert ok
10:16 Manikandan joined #gluster-dev
10:35 gem joined #gluster-dev
11:03 spalai left #gluster-dev
11:19 kdhananjay joined #gluster-dev
11:25 kdhananjay joined #gluster-dev
11:33 sankarshan joined #gluster-dev
11:35 kdhananjay1 joined #gluster-dev
11:36 zhangjn joined #gluster-dev
11:36 zhangjn joined #gluster-dev
11:37 EinstCrazy joined #gluster-dev
11:38 EinstCrazy joined #gluster-dev
11:42 Manikandan joined #gluster-dev
11:43 kdhananjay joined #gluster-dev
11:48 rafi joined #gluster-dev
11:48 kdhananjay joined #gluster-dev
11:49 kdhananjay joined #gluster-dev
12:04 lpabon joined #gluster-dev
12:07 vmallika joined #gluster-dev
12:12 kotreshhr left #gluster-dev
12:12 kotreshhr joined #gluster-dev
12:13 kotreshhr left #gluster-dev
12:19 ppai joined #gluster-dev
12:22 skoduri joined #gluster-dev
12:23 atalur_ pranithk, ping
12:23 glusterbot atalur_: Please don't naked ping. http://blogs.gnome.org/markmc/2014/02/20/naked-pings/
12:29 shubhendu joined #gluster-dev
12:31 overclk joined #gluster-dev
12:31 Manikandan joined #gluster-dev
12:42 shyam joined #gluster-dev
12:58 ppai joined #gluster-dev
13:10 hagarth joined #gluster-dev
13:16 mohan_ joined #gluster-dev
13:34 shyam joined #gluster-dev
13:36 rafi joined #gluster-dev
13:40 ppai joined #gluster-dev
13:45 kkeithley /usr/{libexec,sbin}/gfind_missing_files?  Which subpackage do they belong to? geo-rep?
13:45 hgowtham_ joined #gluster-dev
13:46 csim kkeithley: I am deploying freeipa which may impact your account on the download server (as "may not work"), would it be the right week to do, or next week would be better ?
13:46 csim (I am for now following a long honored tradition of mad scientist of doing things on my own account, but I rather ask before)
13:46 kkeithley lol
13:47 kkeithley well, we have the 3.7.6 release in progress.  But do it. If it breaks then we'll cross that bridge
13:47 csim oh, i rather do it after then
13:47 csim because for now, it kinda do not work :)
13:47 kkeithley oh
13:47 kkeithley okay
13:48 csim but I will make it work like today or tomorow, so I rather not want to remove local account and switch when you use it
13:49 kkeithley so, I'd actually rather have my account match what it is on every other system, i.e. kkeithle.
13:50 csim yeah, i can do that then
13:50 kkeithley currently on download.g.o it's kkeithley.  If you can, if you want to, leave kkeithley for now and I'll use that for now.
13:50 kkeithley Does that work?
13:50 csim yeah, it would work
13:50 kkeithley kewl
13:51 kkeithley let's do that then
13:52 skoduri joined #gluster-dev
13:52 overclk joined #gluster-dev
14:01 ppai joined #gluster-dev
14:03 kkeithley usr/{libexec,sbin}/gfind_missing_files?  Which subpackage do they belong to? geo-rep?
14:04 kkeithley the in-tree glusterfs.spec(.in) has them immediately following the geo-rep sub-package, but outside the %if ... %endif.   In Fedora's dist-git spec they are part of the main package.  Where do they really belong?
14:07 nbalacha joined #gluster-dev
14:11 ira joined #gluster-dev
14:19 mchangir joined #gluster-dev
14:26 ndevos kkeithley: ask your question again, I think mchangir knows the answer
14:28 kkeithley usr/{libexec,sbin}/gfind_missing_files?  Which subpackage do they belong to? geo-rep?
14:28 kkeithley the in-tree glusterfs.spec(.in) has them immediately following the geo-rep sub-package, but outside the %if ... %endif.   In Fedora's dist-git spec they are part of the main package.  Where do they really belong?
14:29 shubhendu joined #gluster-dev
14:30 ndevos kkeithley: :-/ I guess you need to send an email to gluster-devel about it
14:30 rafi joined #gluster-dev
14:32 kkeithley no doubt
14:38 overclk_ joined #gluster-dev
14:41 rafi joined #gluster-dev
14:41 ira joined #gluster-dev
14:42 rafi1 joined #gluster-dev
14:44 shubhendu joined #gluster-dev
14:50 hgowtham_ joined #gluster-dev
15:01 itisravi joined #gluster-dev
15:08 pranithk xavih: Are you there?
15:11 pranithk xavih: I think I found the other problem. The problem it is seeing is the following: 1) when all the bricks are up, it does lock + xattrop before op and figures out all the bricks are good. 2) By the time second operation starts brick-0 is down. Now lock->good_mask will always have the '0' bit as long as the operations are happening on it. because: "lock->good_mask &= ~fop->mask | fop->remaining" fop->mask doesn't have '0' th bit. 3) When it comes time
15:12 pranithk xavih: Let me also give you pm, just in case this scrolls over on your IRC
15:14 poornimag joined #gluster-dev
15:18 Manikandan joined #gluster-dev
15:22 sbonazzo joined #gluster-dev
15:23 sbonazzo hi, can you please create fedora 23 repo for http://download.gluster.org/pub/gluster/glusterfs/LATEST/Fedora/ ?
15:23 ndevos ah kkeithley, I wanted to ask if you can do that ^? I dont think I have access to d.g.o anymore
15:23 sbonazzo ndevos: thanks :-)
15:24 ndevos sbonazzo: no problem :)
15:25 kkeithley I'm in the middle of building. When everthing is done then the LATEST symlink will move.  And you can get 3.7.6 for Fedora 23 from dnf (or yum). Just enable the Updates-Testing repo
15:27 sbonazzo kkeithley: can you please create the fedora 23 repo anyway, also if empty?
15:28 kkeithley ???  We don't usually do that. Is there some reason for wanting this?
15:30 kkeithley s/We don't usually do that./We don't usually create empty repo dirs/
15:31 kkeithley Is there a reason for it?
15:40 sbonazzo kkeithley: CI jobs failing :-) http://jenkins.ovirt.org/view/FC23%20jobs/job/repos_master_check-closure_merged/DISTRIBUTION=fedora23/772/console
15:42 sbonazzo kkeithley: also upgrading fc22 to fc23 with gluster repo enabled will cause issues
15:42 sbonazzo kkeithley: with repository not found
15:43 sbonazzo kkeithley: (well, dnf skip it automatically, but still...)
15:45 kkeithley yes....   well, I still don't understand. If I create an empty f23 repo for 3.7.x on download.gluster.org, then what?  You should be getting 3.7.x from Fedora, not from download.gluster.org repos.
15:50 kkeithley sometimes I'm just slow
15:53 overclk joined #gluster-dev
15:54 sbonazzo kkeithley: you're right, it just won't fail due to missing repository
15:56 kkeithley I see. I think.
16:02 wushudoin joined #gluster-dev
16:15 overclk joined #gluster-dev
16:16 kanagaraj joined #gluster-dev
16:21 jiffin joined #gluster-dev
16:22 ira joined #gluster-dev
16:24 kotreshhr joined #gluster-dev
17:08 overclk joined #gluster-dev
17:17 rafi joined #gluster-dev
17:32 rafi joined #gluster-dev
17:36 kotreshhr left #gluster-dev
18:16 vimal joined #gluster-dev
18:36 rafi joined #gluster-dev
18:42 hagarth joined #gluster-dev
18:47 csim can I reboot the jenkins master ?
18:51 csim ok, so I did
18:57 shaunm joined #gluster-dev
19:32 EinstCrazy joined #gluster-dev
19:49 RedW joined #gluster-dev
20:06 kkeithley sbonazzo: when you say "empty repo" what is it you really want.  just an empty top-level directory .../3.7/3.7.6/Fedora/fedora-23/ ?  Or you need .../3.7/3.7.6/Fedora/fedora-23/{i686,noarch,SRPMS,x86_64}/repodata/*,  just with no actual RPMs there?
20:08 kkeithley ugh, I'm probably not going to see the reply because I'll be wfh tomorrow.  ping me at kkeithley_ when I come on-line tomorrow
20:10 rafi joined #gluster-dev
20:10 kkeithley ndevos: ^^^  by any chance do you know what he needs?
20:11 ndevos kkeithley: yeah, just an empty repository, no rpms
20:11 ndevos kkeithley: just so that existing .repo files do not give any errors anymore
20:12 kkeithley so I need to create a bunch of empty dirs and run `createrepo .` in them
20:30 * kkeithley wonders why we have to jump through hoops just because their CI is broken.
20:38 ndevos kkeithley: I think its our .repo file that is not correct for certain fedora releases? an empty repo is probably nicer for all of our users
20:42 csim yeah, if you have auto detection of the release, it would break
20:56 ira joined #gluster-dev
21:05 csim joined #gluster-dev
21:09 lpabon joined #gluster-dev
21:25 cholcombe joined #gluster-dev
22:32 poornimag joined #gluster-dev
23:00 shyam joined #gluster-dev
23:07 amye joined #gluster-dev

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