Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2015-07-15

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

All times shown according to UTC.

Time Nick Message
00:33 mikedep333 joined #gluster-dev
00:35 vmallika joined #gluster-dev
00:48 sankarshan_ joined #gluster-dev
00:53 topshare joined #gluster-dev
00:59 dlambrig_ joined #gluster-dev
01:11 dlambrig_ left #gluster-dev
01:23 topshare joined #gluster-dev
01:47 ilbot3 joined #gluster-dev
01:47 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/
01:51 kdhananjay joined #gluster-dev
02:05 shyam joined #gluster-dev
02:36 aravindavk joined #gluster-dev
03:11 craigcabrey joined #gluster-dev
03:30 topshare joined #gluster-dev
03:34 nishanth joined #gluster-dev
03:37 atinm joined #gluster-dev
03:40 overclk joined #gluster-dev
03:41 badone joined #gluster-dev
03:52 nishanth joined #gluster-dev
03:57 ppai joined #gluster-dev
03:57 kanagaraj joined #gluster-dev
04:02 shubhendu joined #gluster-dev
04:10 sakshi joined #gluster-dev
04:13 nbalacha joined #gluster-dev
04:16 _ndevos joined #gluster-dev
04:17 itisravi joined #gluster-dev
04:18 overclk joined #gluster-dev
04:31 topshare joined #gluster-dev
04:41 spandit joined #gluster-dev
04:41 overclk joined #gluster-dev
04:43 gem joined #gluster-dev
04:44 ndarshan joined #gluster-dev
04:45 overclk_ joined #gluster-dev
05:00 jiffin joined #gluster-dev
05:13 rafi joined #gluster-dev
05:16 hchiramm joined #gluster-dev
05:16 pppp joined #gluster-dev
05:18 ggarg joined #gluster-dev
05:31 hgowtham joined #gluster-dev
05:31 Manikandan joined #gluster-dev
05:31 vikumar joined #gluster-dev
05:33 rafi joined #gluster-dev
05:34 kdhananjay joined #gluster-dev
05:35 kdhananjay joined #gluster-dev
05:37 ashiq joined #gluster-dev
05:42 Bhaskarakiran joined #gluster-dev
05:46 soumya joined #gluster-dev
05:51 topshare joined #gluster-dev
05:53 ashish joined #gluster-dev
05:57 jiffin1 joined #gluster-dev
05:58 Saravana_ joined #gluster-dev
06:00 soumya joined #gluster-dev
06:03 pranithk joined #gluster-dev
06:04 pranithk _ndevos: did you get a chance to review http://review.gluster.com/6459
06:09 deepakcs joined #gluster-dev
06:13 hagarth pranithk: I think _ndevos won't be available today as he is traveling
06:15 pppp joined #gluster-dev
06:16 kdhananjay joined #gluster-dev
06:22 ndevos hagarth: indeed, but I'll first get breakfast ;-)
06:22 ndevos pranithk: I'll have a look at the change later today, and will post comments tonight or tomorrow morning
06:32 overclk joined #gluster-dev
06:33 topshare joined #gluster-dev
06:37 ggarg joined #gluster-dev
06:42 jiffin joined #gluster-dev
06:58 saurabh_ joined #gluster-dev
07:07 topshare_ joined #gluster-dev
07:33 itisravi_ joined #gluster-dev
07:33 itisravi_ joined #gluster-dev
07:33 vmallika joined #gluster-dev
07:35 ggarg joined #gluster-dev
07:37 topshare joined #gluster-dev
07:40 ndevos hagarth, atinm: because I'll be in the train during todays meeting, I can not guarantee my availability, could you arrange someone to hold meeting?
07:40 ndevos thanks!
07:41 * ndevos disconnects and might be back (much?) later
07:53 pranithk xavih: DId you get a chance to look at http://review.gluster.org/11670
07:54 pranithk xavih: I believe that should be good enough?
07:54 pranithk xavih: http://review.gluster.org/6459 change is needed for this to work...
07:54 hagarth ndevos: sure
07:55 kdhananjay1 joined #gluster-dev
07:56 topshare joined #gluster-dev
08:04 gem joined #gluster-dev
08:05 ggarg joined #gluster-dev
08:09 xavih pranithk: unless you add me as a reviewer it's very unlikely that I see the patch...
08:09 xavih pranithk: I'll take a look
08:18 pranithk xavih: Damn it, I always forget :-(, sorry
08:25 pranithk kdhananjay: wangzhen reuploaded your change for sharding. I think there is on hjhj at the end of your mail address on the commit message. Check that once
08:25 pranithk kdhananjay: may be some accident
08:25 kdhananjay pranithk: Just saw that. Only the commit message seems to have been modified
08:25 kdhananjay pranithk: To be precise, the sign-off.
08:26 pranithk kdhananjay: yeah
08:27 ira joined #gluster-dev
08:27 nbalacha joined #gluster-dev
08:29 itisravi joined #gluster-dev
08:33 kdhananjay pranithk: Updated the commit message.
08:33 pranithk kdhananjay: cool, thanks
08:34 pranithk kdhananjay: regression will run again :-/
08:34 pranithk kdhananjay: I gave +2
08:34 kdhananjay pranithk: I know!
08:34 topshare joined #gluster-dev
08:35 krishnan_p joined #gluster-dev
08:35 xavih pranithk: I've reviewed the patch. It doesn't fully solve all possible races yet
08:36 xavih pranithk: there are other races with timers not addressed by your patch, but they would need more modifications to patch 6459
08:37 pranithk xavih: Oh, what kind of races?
08:39 xavih pranithk: I think the explanation in your commit message is not correct for forced unlocks
08:39 xavih pranithk: but let me think a bit more...
08:40 xavih pranithk: yes, I think my comment is right. There's still one race
08:40 xavih pranithk: timer implementation still has races for termination functions
08:40 pranithk xavih: yes yes I only thought of race between unlock force and new lock
08:40 pranithk xavih: but forgot about force unlock and timer
08:41 overclk joined #gluster-dev
08:41 pranithk xavih: I will change unlock_timer_del with if (call_cancel is unsuccessful) then it shouldn't do the things in that if block
08:41 xavih pranithk: other issues with timers: for example the timer created to bring volume up if not all subvolumes have come online has a race if the volume if brought down very fast
08:42 xavih pranithk: that does not solve the problem: if the timer has already been fired, neither the callback nor the forced unlock will really unlock the lock
08:43 pranithk xavih: no no, timer will unlock
08:43 pranithk xavih: oh damn
08:43 pranithk xavih: I got it
08:43 xavih pranithk: how ?
08:43 pranithk xavih: interesting problem :-)
08:43 xavih pranithk: only solution is separate functions for callback and forced unlock
08:44 pranithk xavih: thinking...
08:44 xavih pranithk: or rework timer implementation :P
08:44 pranithk xavih: :-)
08:45 pranithk xavih: I am under lot of pressure for this fix actually, new timer implementation now will not be accepted now, but definitely something to work for in future
08:45 shubhendu joined #gluster-dev
08:45 xavih pranithk: I know :)
08:47 xavih pranithk: but be aware that there are other issues not solved yet (at least one with timer cancellations on fini() )
08:47 pranithk xavih: yes yes definitely
08:47 xavih pranithk: however this one is much less probable and less critical
08:48 pranithk xavih: yes
08:49 xavih pranithk: do you want to handle the forced unlock/callback race in this patch or do you want to leave it for a future patch ?
08:50 xavih pranithk: I think it should be solved now to avoid failures on graph switch...
08:50 pranithk xavih: I want to fix it off. One more bug at this time, I will be screwed :-)
08:55 ndarshan joined #gluster-dev
08:57 jiffin1 joined #gluster-dev
09:03 pranithk xavih: force_unlocks seems to have race?
09:03 pranithk xavih: there is no guarantee to prevent fop being destroyed?
09:04 xavih pranithk: the guarantee is that one only one thread unlocks the lock
09:05 pranithk xavih: why?
09:05 xavih pranithk: if two threads do it, it's bad
09:05 xavih pranithk: I don't understand
09:05 pranithk xavih: okay, check the following code ec_force_unlocks
09:06 pranithk xavih: while loop is racy, another thread can go ahead and free the fop
09:07 pranithk xavih: I mean force_unlocks will try to call ec_unlock_force, meanwhile someother thread can go ahead and free this fop...
09:08 sakshi joined #gluster-dev
09:09 pranithk xavih: Worst case it will wait 1 second... thats all. I am thinking it is better to remove the ec->pending_fops handling in ec_force_unlocks
09:10 xavih pranithk: I'm thinking...
09:15 pranithk xavih: Let me tell you the case I have in mind. thread-1: executing ec_force_unlocks(), going to execute ec_unlock_force on fop1, thread-2: executing ec_handle_last_pending_fop_completion on same fop1
09:15 pranithk xavih: as soon as thread-1 relinquishes the lock, thread-2 will go ahead and free up fop1...
09:17 xavih pranithk: I think there's a problem, but I don't clearly see it...
09:17 pranithk xavih: example above doesn't help?
09:18 ira joined #gluster-dev
09:19 xavih pranithk: If thread 2 is executing ec_handle_last_pending_fop_completion, this means that the timer has already expired, and fop fully unlocked, right ?
09:19 xavih pranithk: in this case, whatever thread 1 does will be ok (the timer will be already cancelled)
09:19 pranithk xavih: yes, but it will still be in pending fops
09:20 pranithk xavih: thread-2 can go ahead and freeup before thread-1 can access 'link' structure?
09:20 xavih pranithk: yes, you are right
09:21 pranithk xavih: Waiting for 1 second is fine IMO
09:21 xavih pranithk: If waiting up to 1 second is not a problem, it's ok for me
09:21 pranithk xavih: new fops are anyway not going to go for delayed locks because we check for ec->shutdown before starting delay timer
09:22 xavih pranithk: yes
09:22 rjoseph joined #gluster-dev
09:22 pranithk xavih: cool then I will remove the list processing part in that function. Even this race will be gone :-)
09:22 xavih pranithk: great :)
09:24 hagarth joined #gluster-dev
09:27 ggarg joined #gluster-dev
09:28 overclk joined #gluster-dev
09:28 atinm joined #gluster-dev
09:31 overclk joined #gluster-dev
09:33 kdhananjay joined #gluster-dev
09:37 overclk joined #gluster-dev
09:38 hgowtham joined #gluster-dev
09:40 nbalacha joined #gluster-dev
09:53 soumya joined #gluster-dev
10:04 pranithk xavih: resent them :-)
10:04 pranithk xavih: Can you believe that this race happens every 40 minutes on Bhaskarakiran's setup?
10:04 ndarshan joined #gluster-dev
10:05 shubhendu joined #gluster-dev
10:15 overclk joined #gluster-dev
10:19 kshlm joined #gluster-dev
10:25 atinm joined #gluster-dev
10:28 ggarg joined #gluster-dev
10:30 csim so, I managed to have 1 VM running on the ci host
10:30 csim now looking at making sure it doesn't break on reboot, and that jenkins is setup
10:30 csim (and I was out for a few days, due to national holiday, etc, etc )
10:31 nbalacha joined #gluster-dev
10:45 kdhananjay joined #gluster-dev
10:50 hagarth atinm: ping, will you be able to run the community meeting today?
10:57 aravindavk joined #gluster-dev
11:17 soumya joined #gluster-dev
11:17 krishnan_p joined #gluster-dev
11:18 pppp joined #gluster-dev
11:18 shubhendu joined #gluster-dev
11:25 overclk joined #gluster-dev
11:29 rafi1 joined #gluster-dev
11:31 suliba_ joined #gluster-dev
11:32 kanagaraj_ joined #gluster-dev
11:32 shaunm__ joined #gluster-dev
11:33 kdhananjay joined #gluster-dev
11:37 rafi1 joined #gluster-dev
11:37 rafi joined #gluster-dev
11:37 kanagaraj_ joined #gluster-dev
11:37 kdhananjay joined #gluster-dev
11:37 rafi joined #gluster-dev
11:49 bfoster joined #gluster-dev
11:59 raghu joined #gluster-dev
11:59 topshare joined #gluster-dev
12:00 dlambrig_ joined #gluster-dev
12:00 kshlm Gluster community meeting is starting in #gluster-meeting now.
12:04 jdarcy joined #gluster-dev
12:07 shubhendu joined #gluster-dev
12:09 soumya joined #gluster-dev
12:12 topshare joined #gluster-dev
12:13 topshare joined #gluster-dev
12:13 jrm16020 joined #gluster-dev
12:17 ashiq joined #gluster-dev
12:18 surabhi joined #gluster-dev
12:19 krishnan_p joined #gluster-dev
12:30 lpabon joined #gluster-dev
12:35 ashiq joined #gluster-dev
12:37 Manikandan sakshi++, thanks:)
12:37 glusterbot Manikandan: sakshi's karma is now 3
12:43 shubhendu joined #gluster-dev
12:43 aravindavk joined #gluster-dev
12:44 topshare joined #gluster-dev
12:46 ppai joined #gluster-dev
12:59 ashiq joined #gluster-dev
13:00 Manikandan joined #gluster-dev
13:12 kkeithley hchiramm++
13:12 glusterbot kkeithley: hchiramm's karma is now 54
13:27 Manikandan joined #gluster-dev
13:34 kshlm ndevos, you around?
13:36 shyam joined #gluster-dev
13:36 hagarth kshlm: ndevos is mostly traveling today
13:36 ndevos kshlm: a little... on/off connection
13:37 kshlm hagarth, okay. I wanted to ask how he put next weeks meeting invite in the meeting summary he sent.
13:37 hagarth ndevos: you mostly prove me wrong ;)
13:37 hagarth kshlm: I think he knows that best
13:37 ndevos hagarth: lol, yes thats the 2nd time today :)
13:38 kshlm ndevos, you're here. I thought it was hagarth replying again. Both of you have the same colour in my IRC client.
13:38 ndevos kshlm: I used thinderbird, created an event and copied into a text file
13:39 kshlm and attached the text file as an ics file?
13:39 ndevos kshlm: you'll get a raw ics formatted uglyness, and that can be saved as a .ics file
13:39 ndevos kshlm: yes, like that
13:40 kshlm Okay. Thanks.
13:40 kshlm ndevos++
13:40 glusterbot kshlm: ndevos's karma is now 178
13:40 ndevos kshlm: and thunderbird/lightning has timezone called UTC, thats awesome!
13:42 ndevos kshlm: in case you're looking for the timezone, it's in the event's menu-bar under Opptions -> "show timezones"
13:43 ndevos each event can happen in its own timezone, its not a generic option for lightning itself
13:46 ashiq joined #gluster-dev
13:49 kanagaraj_ joined #gluster-dev
14:03 shubhendu joined #gluster-dev
14:06 hagarth joined #gluster-dev
14:10 csim kkeithley: you have the required access on jenkins to add a server ? I suspect I can force my way using the CLI but I prefer to be cautious and ask for this time
14:10 csim ( as "let's be bold" didn't work well in the past )
14:10 pousley joined #gluster-dev
14:11 csim kkeithley: the jenkins fedora would be on 192.168.123.177 from build.gluster
14:11 nbalacha joined #gluster-dev
14:14 kkeithley I never remember. Let me try
14:15 kkeithley if access means "can sign on to build.gluster.org and have sudo" then yes, I have access
14:15 kkeithley csim: ^^^
14:16 csim kkeithley: more access to the web interface, I do have that access
14:16 csim but all I find on jenkins doc start by "login on the web interface"
14:16 csim but ok, i will find a way :)
14:17 shyam joined #gluster-dev
14:17 kkeithley I have an account, I'm signed on
14:18 kkeithley Manage nodes?
14:19 csim yep
14:19 csim or rather, can you just give me a account and I can do it myself ?
14:19 kkeithley okay, let me find where to do that
14:19 csim or just give your password and the credit card as well :p
14:20 kkeithley lol
14:21 vimal joined #gluster-dev
14:22 csim I am impressed that "adding user" do not give the result I want nor "user create"
14:23 csim https://stackoverflow.com/questions/1205​6851/add-users-to-jenkins-with-allow-use​rs-to-sign-up-disabled/12365640#12365640
14:24 kkeithley I'm not finding a place in the GUI to add users
14:24 topshare joined #gluster-dev
14:25 csim lovely
14:25 csim ok, I will try to find a way or summon justin
14:26 kkeithley configure global security. I looks like it uses the Unix user/group database. If you have a unix account on the box I think you can sign into the gui using your unix passwd
14:27 csim oh
14:27 csim ok, would make sense
14:27 csim ( of course, I use a ssh key and I erased the password from memory using potato based distrilled chemical, so it will be hard )
14:28 kkeithley but you can ssh in and set a new passwd, right?
14:29 csim yup
14:29 kkeithley potato-based distilled chemical, a.k.a. vodka?
14:29 csim kkeithley: yup :)
14:29 csim so, I am not gonna type my password on cleartext...
14:29 csim yeah for yakshaving
14:30 jbautista- joined #gluster-dev
14:30 kkeithley well, use a throw-away passwd. The jenkins UI isn't https.
14:30 topshare joined #gluster-dev
14:31 kkeithley The jenkins UI isn't https  :-(
14:31 csim I can take care of that I guess
14:31 csim or use a ssh tunnel
14:31 topshare joined #gluster-dev
14:31 csim but tea time
14:32 kkeithley yes, it would be good to get to a newer jenkins, and with https.
14:34 csim https could be added
14:34 csim just put a proxy between jenkins and the world :)
14:34 csim https can be done, using a self signed cert, and then get a proper one
14:35 csim newer jenkins, I was planning to switch to the rpm in copr but I really need to spend more time on jenkins before moving
14:36 kkeithley We may have a cert for review.gluster.org. Maybe johnmark knows.
14:37 ira csim: If you know enough to do the update... why learn the older one ;)
14:37 ira </evil>
14:38 ndevos csim: ssh to build.gluster.org is (or was?) only allowed with ssh-keys, for all I know, Jenkins is the only user of the insecure password?
14:48 shyam joined #gluster-dev
14:51 jobewan joined #gluster-dev
14:54 shaunm_ joined #gluster-dev
14:56 overclk joined #gluster-dev
14:59 topshare joined #gluster-dev
15:01 hchiramm_home joined #gluster-dev
15:20 topshare joined #gluster-dev
15:22 kdhananjay joined #gluster-dev
15:58 hagarth joined #gluster-dev
16:17 soumya joined #gluster-dev
16:20 dlambrig_ joined #gluster-dev
16:54 craigcabrey joined #gluster-dev
16:54 csim yeah, so jenkins, even with https, redirect to http after
16:57 firemanxbr joined #gluster-dev
17:05 overclk joined #gluster-dev
17:17 pppp joined #gluster-dev
17:31 jiffin joined #gluster-dev
17:32 gem joined #gluster-dev
17:34 gem joined #gluster-dev
17:36 ggarg joined #gluster-dev
17:37 gem joined #gluster-dev
17:39 gem joined #gluster-dev
17:41 gem joined #gluster-dev
17:59 gem joined #gluster-dev
19:06 jdarcy joined #gluster-dev
20:52 badone joined #gluster-dev
20:59 dlambrig1 left #gluster-dev
22:02 shyam joined #gluster-dev
22:23 hchiramm_home joined #gluster-dev
22:25 topshare joined #gluster-dev
22:46 nishanth joined #gluster-dev
23:09 topshare joined #gluster-dev
23:17 topshare joined #gluster-dev
23:42 lpabon joined #gluster-dev
23:44 topshare joined #gluster-dev
23:46 topshare_ joined #gluster-dev
23:54 chinita70 joined #gluster-dev

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