--- Log opened do jul 24 00:00:05 2014 01:16 -!- apuimedo (antoni): has joined #vdsm 01:20 -!- nsoffer: has quit [Read error: Connection reset by peer] 01:40 < osvoboda> apuimedo: I just told danken I could just squash parts of http://gerrit.ovirt.org/#/c/30532/16 to the backported BOOTPROTO patch to verify it. 01:41 < apuimedo> osvoboda: I'm not really sure we need to backport it 01:44 < osvoboda> apuimedo: Maybe it is not strictly needed but it would show that under normal conditions (not when reconfiguring the network) all the bootproto values are as expected. 01:46 < apuimedo> osvoboda: I, for one, I'm content with having that in master 01:46 < osvoboda> :-) 01:46 < apuimedo> let's first get it on master 01:47 < apuimedo> and then maybe for safety we can merge it as a separate patch (not squashing) with instead of Bug-Url Related-To 01:47 < osvoboda> On a related note, I reeeally want to rename bootproto4 to dhcpv4 and make it boolean. 01:47 < osvoboda> apuimedo: We could do that, yep. 01:52 < osvoboda> apuimedo: The test that danken asked for (that even after reconfiguring the network with DHCP, dhclient is "still"/again there) could either use pgrep and a PID file or use a pre-made lease file with current/outdated leases (both a result of dhclient -x). 01:52 < apuimedo> I'm doing it now ;-) 01:52 < apuimedo> it uses /proc/pid/cmdline 01:52 < osvoboda> Do you ever sleep? :-D 01:52 < osvoboda> apuimedo: Good idea! 01:52 < apuimedo> not enough 01:57 -!- bala (purple): has joined #vdsm 01:59 < osvoboda> I, too, would like to have dhclient take over cleanly. Did it really just exit with exitcode 2 when it got an already configured interface? I guess it can never be sure if there ever was another dhclient (with leases/pidfiles elsewhere) or if it was manually configured. If so, removing the address is the way to go. And hopefully reliable :-) 02:02 < apuimedo> it just exits 02:03 < apuimedo> with 2 02:03 < apuimedo> the check for existing leases is afterwards 02:04 < osvoboda> So if there is no address, can we count on it checking that the lease is valid? 02:04 < apuimedo> it checks if there is a valid lease 02:04 < apuimedo> and if there is not 02:04 < apuimedo> it sends a discovery 02:05 < osvoboda> All good. 02:06 < osvoboda> If it indeed renews the valid lease in time, as well. 02:06 < apuimedo> it daemonizes 02:06 < apuimedo> so it should 02:15 < apuimedo> osvoboda: patch updated with the functional test 02:15 < apuimedo> ;-) 02:17 < osvoboda> apuimedo: Nice one :-) 02:18 < apuimedo> simpler than I thought it would be 02:19 < osvoboda> Mine is broken but it’s too late now. I just uploaded a version where:['BOOTPROTO'] is checked again only the bridged case, the rest of failures is for tomorrow. 02:28 -!- gpadgett: has quit [Quit: Leaving] 02:29 < osvoboda> Oooh, mismatched parameters for dnsmasqDhcp(). Let's see. 02:30 -!- shaohef: has quit [Quit: Leaving.] 02:46 -!- #vdsm dyasny: has quit [Ping timeout: 255 seconds] 02:50 < osvoboda> Good night! 02:51 -!- osvoboda: has quit [Quit: ChatZilla 0.9.90.1 [Firefox 31.0/20140722064054]] 02:53 -!- apuimedo: has quit [Ping timeout: 255 seconds] 04:01 -!- shaohef (shaohef): has joined #vdsm 04:34 -!- shaohef: has quit [Ping timeout: 260 seconds] 04:34 -!- shaohef (shaohef): has joined #vdsm 04:45 -!- adahms_ (Andrew Dahms): has joined #vdsm 04:48 -!- adahms: has quit [Ping timeout: 240 seconds] 04:48 -!- adahms_ is now known as adahms-mtg 04:51 -!- shaohef: has quit [Ping timeout: 240 seconds] 05:00 -!- mliu (purple): has joined #vdsm 05:00 -!- mliu: has left #vdsm [] 05:02 -!- shaohef (shaohef): has joined #vdsm 05:38 -!- adahms-mtg: has quit [Ping timeout: 240 seconds] 05:51 -!- adahms-mtg (Andrew Dahms): has joined #vdsm 06:45 -!- shaohef: has quit [Ping timeout: 240 seconds] 07:08 -!- shaohef (shaohef): has joined #vdsm 07:31 -!- adahms-mtg: has quit [Ping timeout: 255 seconds] 07:36 -!- sbonazzo (purple): has joined #vdsm 07:44 -!- adahms-mtg (Andrew Dahms): has joined #vdsm 07:45 -!- aravindavk (Aravinda): has joined #vdsm 07:49 -!- #vdsm ybronhei: has quit [Quit: Leaving.] 08:06 -!- evilissimo|afk is now known as evilissimo 08:07 -!- kobi (Kobi Ianko): has joined #vdsm 08:08 -!- adahms-mtg: has quit [Ping timeout: 245 seconds] 08:21 -!- adahms-mtg (Andrew Dahms): has joined #vdsm 08:28 -!- mskrivanek_away is now known as mskrivanek 08:36 -!- ybronhei (purple): has joined #vdsm 08:45 -!- phoracek (phoracek): has joined #vdsm 08:46 -!- adahms-mtg: has quit [Quit: Leaving] 08:47 -!- osvoboda (osvoboda): has joined #vdsm 08:57 -!- mskrivanek is now known as mskrivanek_away 08:57 -!- mskrivanek_away is now known as mskrivanek 09:03 -!- phoracek: has quit [Ping timeout: 260 seconds] 09:04 -!- sbonazzo is now known as sbonazzo|afk 09:04 -!- aravindavk: has quit [Ping timeout: 245 seconds] 09:09 -!- shaohef: has quit [Ping timeout: 250 seconds] 09:16 -!- pkliczew (Piotr Kliczewski): has joined #vdsm 09:19 -!- phoracek (phoracek): has joined #vdsm 09:20 -!- aravindavk (Aravinda): has joined #vdsm 09:27 -!- fromani (Francesco Romani): has joined #vdsm 09:27 -!- bazulay: has quit [Quit: Leaving.] 09:33 -!- phoracek: has quit [Ping timeout: 255 seconds] 09:42 -!- msivak (Martin Sivák): has joined #vdsm 09:46 -!- phoracek (phoracek): has joined #vdsm 09:51 -!- phoracek: has quit [Quit: WeeChat 0.4.3] 09:51 -!- phoracek (phoracek): has joined #vdsm 09:54 -!- aravindavk: has quit [Ping timeout: 250 seconds] 10:02 -!- moolit (purple): has joined #vdsm 10:04 < ybronhei> danken: here? 10:04 -!- phoracek: has quit [Quit: WeeChat 0.4.3] 10:04 -!- moolit1 (purple): has joined #vdsm 10:06 -!- shaohef (shaohef): has joined #vdsm 10:06 -!- sbonazzo|afk is now known as sbonazzo 10:06 -!- moolit: has quit [Ping timeout: 240 seconds] 10:07 <@danken> ybronhei: yep 10:07 -!- aravindavk (Aravinda): has joined #vdsm 10:08 <@danken> ybronhei: please get the rmFile patch into ovirt-3.5 10:08 < ybronhei> danken: how to check the tag permissions? 10:09 < ybronhei> danken: what patch? 10:10 -!- apuimedo (antoni): has joined #vdsm 10:15 -!- fabiand (Fabian Deutsch): has joined #vdsm 10:17 -!- derez_ (Daniel Erez): has joined #vdsm 10:17 -!- derez_: has quit [Client Quit] 10:20 <@danken> ybronhei: http://gerrit.ovirt.org/30545 10:21 <@danken> ybronhei: check it only when we need to tag. 10:21 <@danken> such as after pushing the forceLink patch 10:22 <@danken> tag -a v4.16.1 10:22 <@danken> but be careful, since we have no permission to delete a wrong flag 10:22 <@danken> *tag 10:22 <@danken> only dcaro can do that. 10:23 < ybronhei> danken: that's what i thought 10:27 < ybronhei> hash-of-ovirt-3.5-barnch means last commit I guess, right? 10:27 < ybronhei> danken: ^ ? 10:33 < ybronhei> pkliczew: what's with http://gerrit.ovirt.org/#/c/30033/ ? 10:33 < ybronhei> before i tag 3.5 .. 10:33 < pkliczew> ybronhei: let me check 10:34 -!- nsoffer (Nir Soffer): has joined #vdsm 10:35 < pkliczew> ybronhei: Saggi asked to revert 6c2892a2, in my option it should be part of 3.5 10:35 < pkliczew> ybronhei: please wait with this patch for heart beat 10:35 < pkliczew> ybronhei: I will let you know when 10:36 <@danken> ybronhei: yeah, the tip of the branch. unless you want to tag something a bit older 10:37 <@danken> ybronhei: but you are responsible to make the version sequential (increase only micro part) 10:39 <@danken> ybronhei: btw, once osvoboda verifies http://gerrit.ovirt.org/30646 we'd need it in the stable branch. 10:40 < ybronhei> yes, afaiu we don't need tag right now. when we'll post new build i'll tag 10:40 < ybronhei> can you take - http://gerrit.ovirt.org/#/c/30667/ ? 10:40 < ybronhei> its annoying me 10:50 -!- acanan (Aharon Canan): has joined #vdsm 11:02 <@danken> ybronhei: and I actually like more info 11:02 < ybronhei> what more info you want 11:02 <@danken> could you better explain what is that evil text, when is it shown 11:03 < ybronhei> oh come on 11:03 <@danken> and why is it unhelpful 11:03 <@danken> I really do not understand 11:03 <@danken> ybronhei: I'm not trying to be nasty, I just love copious logs 11:04 < ybronhei> during rpm installation ? 11:04 < ybronhei> wait, ill have it in a sec 11:04 < ybronhei> http://fpaste.org/120406/92674140/ 11:05 < ybronhei> are those important? "Running configure..." "Done configuring modules to VDSM." 11:12 <@danken> ybronhei: I don't find them offensive. but that's the positive flow. the problem is reporting errors 11:12 < ybronhei> during rpm installation we should not print anything to stdout 11:12 < ybronhei> on error you should check your syslog 11:22 <@danken> I know that this is the Fedora rule... 11:22 <@danken> but how would vdsm-tool problems be recognized? 11:40 < ybronhei> what problem can be ? 11:40 < ybronhei> when you'll start vdsmd you will see the problems 11:41 < ybronhei> anyhow when you do "yum upgrade -y " you don't check on each package the errors on the yum log 11:41 < ybronhei> danken: ^ 11:44 -!- hchiramm: has quit [Quit: Leaving] 11:45 -!- hchiramm (Humble Chirammal): has joined #vdsm 11:45 <@danken> ybronhei: there can be all kinds of erros; when vdsm does not start, it makes sense to dig up in the history of yum.log 11:46 <@danken> ybronhei: I can see why you'd like to drop stdout. But why is stderr so troubling? 11:46 < ybronhei> i dont agree with that much . if the configure part raise errors, you should check it in syslog 11:46 <@danken> errors there should be rare and important 11:47 <@danken> ybronhei: would they get into syslog? how? 11:49 < ybronhei> i can also redirect exception to syslog 11:50 < ybronhei> execptions, in vdsm-tool 11:50 < ybronhei> all others than UsageError , which could be relevant 11:56 -!- hchiramm: has quit [Ping timeout: 250 seconds] 11:58 -!- shaohef: has quit [Ping timeout: 240 seconds] 12:01 -!- derez_ (Daniel Erez): has joined #vdsm 12:02 -!- danken is now known as danken_mtng 12:03 -!- bazulay (purple): has joined #vdsm 12:10 -!- hchiramm (Humble Chirammal): has joined #vdsm 12:28 < osvoboda> danken: Verified http://gerrit.ovirt.org/#/c/30646/ 12:29 < osvoboda> danken_mtng: After you have finished the meeting, ^^ 13:00 -!- phoracek (phoracek): has joined #vdsm 13:13 <@danken_mtng> osvoboda: thanks 13:13 -!- danken_mtng is now known as danken 13:23 -!- mliu (purple): has joined #vdsm 13:24 -!- mliu: has left #vdsm [] 13:43 -!- aravindavk: has quit [Ping timeout: 240 seconds] 13:45 -!- #vdsm ybronhei: has quit [Ping timeout: 245 seconds] 14:09 -!- bala: has quit [Ping timeout: 250 seconds] 14:16 -!- bala (purple): has joined #vdsm 14:17 -!- ybronhei (purple): has joined #vdsm 14:21 -!- bala: has quit [Ping timeout: 272 seconds] 14:28 < bazulay> ybronhei: ping 14:33 -!- nsoffer: has quit [Remote host closed the connection] 14:34 -!- nsoffer (Nir Soffer): has joined #vdsm 14:37 -!- phoracek: has quit [Quit: WeeChat 0.4.3] 14:41 -!- shaohef (shaohef): has joined #vdsm 14:42 -!- adahms: has quit [Quit: Leaving] 14:48 -!- apahim (Amador Pahim): has joined #vdsm 15:23 -!- bala1 (purple): has joined #vdsm 15:29 -!- bala1: has quit [Quit: Leaving.] 15:29 -!- gpadgett (Greg Padgett): has joined #vdsm 15:32 -!- #vdsm ybronhei: has quit [Ping timeout: 250 seconds] 15:37 -!- ybronhei (purple): has joined #vdsm 15:40 -!- bala (purple): has joined #vdsm 15:42 < apuimedo> danken: osvoboda: My test fails because dhclient -x does not kill the dhclient run by the test 15:42 < apuimedo> as the pid is in a weird place and dhclient code does not find it 15:43 < apuimedo> so then there are two dhclients for a single device 15:43 < osvoboda> apuimedo: So do we have to move the PID file? 15:43 < osvoboda> Or kill dhclient ourselves? 15:44 * apuimedo is checking 15:44 < apuimedo> osvoboda: it would defeat the purpose of the test 15:44 < apuimedo> the test checks if we kill the old one and setup a new one 15:44 < osvoboda> apuimedo: Yeah, not only the test would have to change, I guess. 15:46 -!- orc_emac1 is now known as orc_emac 15:46 < osvoboda> I mean, I assume that normally, dhclient -x works for initscripts-invoked dhclients and that it is just us who put the PID file somewhere dhclient does not look. I am wrong? 15:47 < apuimedo> osvoboda: that's the case 15:47 < apuimedo> we should just fix the test 15:47 < apuimedo> so that the pidfile location is the standard 15:55 -!- mskrivanek is now known as mskrivanek_away 15:56 < osvoboda> apuimedo: I am afraid dhclient -x is not a solution, judging by its code, because by default it only looks at /var/…/dhclient.pid, it does not list a dictionary. 15:56 < apuimedo> osvoboda: dictionary? 15:57 < osvoboda> apuimedo: Heh, in a filesystem sense, you got me :-) 15:57 < apuimedo> in any case I arrived to a similar conclusion. I have decided to use the code I made for the test 15:57 < apuimedo> to list dhclient instances 15:57 < apuimedo> and manually send a kill to all of them 15:58 < osvoboda> apuimedo: Yes, it looks in a single file unless you pass a -pf specification or an env variable. 15:58 < apuimedo> then flush de ip address 15:58 < apuimedo> yes 15:58 < apuimedo> that's exactly why I'm changing the kill code ;-) 15:58 < osvoboda> apuimedo: Yep, there is no better way I guess. 15:58 < apuimedo> osvoboda: and it is not a bad way either 15:58 < osvoboda> apuimedo: Only you leave a pid file behind. 15:58 < apuimedo> considering what '-x' actually does... 15:59 < apuimedo> osvoboda: well, I could even avoid that 15:59 < apuimedo> hell 15:59 < apuimedo> I should 15:59 < apuimedo> I'll use the /proc/num/cmdline 15:59 < apuimedo> parse the pidfile 15:59 < osvoboda> To look for pidfile, right! :-) 15:59 < apuimedo> and then, instead of sending the kill 16:00 < apuimedo> pass the pidfile to a dhclient -x 16:00 -!- mskrivanek_away is now known as mskrivanek 16:00 < apuimedo> what a boring code 16:00 < osvoboda> Well it only sends a SIGTERM, nothing creative. 16:00 -!- pkliczew: has quit [Ping timeout: 240 seconds] 16:01 < apuimedo> osvoboda: but it handles itself the pid files 16:01 < apuimedo> I'm not sure I should touch its pid files 16:01 -!- kobi: has quit [Remote host closed the connection] 16:01 < osvoboda> You should not need to, I think. 16:02 < osvoboda> If it creates the file, it should remove it if you kill it with SIGTERM. 16:02 < osvoboda> But I’ll look :-) 16:02 < apuimedo> true 16:03 < osvoboda> Um, I am not sure yet, I hoped to grep a little bit more… 16:03 < apuimedo> it's in omapip/isclib.c:dhcp_signal_handler 16:05 < apuimedo> mmm 16:05 < apuimedo> I don't see anything relevant there 16:13 < osvoboda> apuimedo: I have not found anything either. Maybe it is better just to try it out ;-) 16:13 < apuimedo> I'm doing that for the last 3 mins 16:13 < apuimedo> :P 16:18 < apuimedo> screw it, I'll just remove the pid file if it exists just like ifdown-eth does 16:20 < osvoboda> And just handle the case when the file is gone already to be sure :-) 16:20 -!- dyasny (Dan Yasny): has joined #vdsm 16:20 < apuimedo> osvoboda: with an except 16:20 < apuimedo> dyasny: long time no see ;-) 16:20 < osvoboda> Yep. 16:22 -!- LaercioMotta (Laercio da Silva Motta): has joined #vdsm 16:22 -!- evilissimo is now known as evilissimo|afk 16:23 < osvoboda> And maybe, if --no-pid file was not used (therefore, the file was used), look for the default if -pf is missing. In cases env var was used to specify the pidfile path, we can check /proc/pid/environ. But nobody uses that ;-) 16:24 < osvoboda> Stupid thing if it really does not delete its pidfile… 16:24 < apuimedo> ;-) 16:25 < osvoboda> We say in Czech that it’s like a cat catching a mouse. 16:25 < osvoboda> Hra na kočku a myš. 16:25 < apuimedo> we say dog and cat 16:38 < osvoboda> Maybe it’s also 'hide and seek' in English. 16:47 -!- #vdsm dyasny: has quit [Ping timeout: 250 seconds] 16:56 < apuimedo> osvoboda: http://gerrit.ovirt.org/#/c/30651/2 16:57 < apuimedo> osvoboda: http://gerrit.ovirt.org/#/c/30651/3 17:02 -!- dyasny (Dan Yasny): has joined #vdsm 17:06 < dyasny> apuimedo, yup, well, I'm trying to stay online again 17:06 < apuimedo> dyasny: ;-) 17:10 -!- osvoboda: has quit [Remote host closed the connection] 17:11 -!- danken: has quit [Ping timeout: 250 seconds] 17:15 -!- derez_: has quit [Ping timeout: 250 seconds] 17:18 -!- osvoboda (osvoboda): has joined #vdsm 17:25 -!- mode/#vdsm: by ChanServ 17:25 -!- danken (purple): has joined #vdsm 17:34 -!- phoracek (phoracek): has joined #vdsm 17:35 < alitke> fromani, Could you please take a peek at http://gerrit.ovirt.org/#/c/30046/ ? It's the live merge change that I need in before I can get active commit working. 17:36 < fromani> alitke: I did earlier, and I'd like to ask a few questions actually. Do you have a few minutes? 17:36 < alitke> absolutely! 17:36 < fromani> alitke: is for my understanding, I'like to make sure I got all the detaisls 17:36 < fromani> details* 17:36 < alitke> yes, definitely. 17:37 -!- apuimedo: has quit [Ping timeout: 255 seconds] 17:38 < alitke> fromani, Would you prefer to discuss here or on the phone? Here is fine with me so others can tune in as well. 17:39 < fromani> alitke: I think here is fine exactly for the reason you said, the channels seems also quiet right now 17:39 < fsimonce> (listening in) 17:39 -!- mskrivanek is now known as mskrivanek_away 17:39 < fromani> let me just refresh the patch 17:39 < alitke> ok, great 17:40 < fromani> ok, so now we'll have these LiveMergeCleanupThread which will sync up the volume chain asynchornously with respect of the creation thread 17:41 < fromani> and they are -supposedly- short lived 17:41 < fsimonce> (yes) 17:41 < alitke> yes. A function will change some state in the VM instance and then we call HSM.imageSyncVolumeChain 17:42 -!- #vdsm dougsland: has quit [Ping timeout: 255 seconds] 17:42 < fromani> fine. I see they also need to interact only with the queryBlockJobs method 17:42 < alitke> yes 17:42 < osvoboda> apuimedo: I have a suggestion to your test. http://gerrit.ovirt.org/#/c/30651/ 17:42 < fromani> ok cool 17:43 < osvoboda> Oh, he’s gone. 17:43 < fromani> as fsimonce already pointed out on gerrit, we don't usually set thread references in the vm.conf 17:43 < fromani> the common way to go is to add fields to the Vm object 17:43 < alitke> Yeah... If it's a deal-breaker I can place them in a dict in the main VM class 17:44 < alitke> was just trying to keep them out of the way 17:44 < alitke> that is an easy change though 17:44 < fromani> indeed to add fields is not the best thing in the world but that's is another story 17:44 < fsimonce> that'd be nice (together with a small comment on the if/elif clause) 17:44 < fromani> alitke: yes please move them away 17:44 < fromani> alitke: moreover I'd like to limit the number of _hidden fields in self.conf 17:45 < alitke> fromani, Fair point. I'll fix that. 17:45 < fromani> alitke: thanks 17:45 < alitke> fsimonce, Yes, glad to add comments to outline the different cases regarding the cleanup thread. 17:46 < fromani> alitke: fsimonce other than that I don't have big remarks yet, just let me digest the concept a bit more. Rest assured is very high on my priority queue. 17:46 < fsimonce> perfect, fromani can you take a closer look at other things as well? I would like to merge it today or tomorrow 17:47 < alitke> fromani, Ok 17:47 < alitke> thanks 17:47 < fromani> fsimonce: sure, please just send me the list so I'll make sure I'll update my TODO 17:47 < fsimonce> let's say that by tomorrow afternoon (italian tz) we need to converge and get it merged 17:47 -!- acanan: has quit [Quit: Leaving] 17:48 < alitke> fromani, I'll update the patch and send you the list. It's two patches 17:48 < fromani> alitke: fine. I'm good with fsimonce's timing 17:48 < alitke> the other one you've already reviewed and it has no significant changes. 17:48 < alitke> great! 17:48 < alitke> thanks guys! 17:49 < fromani> np! 17:49 -!- #vdsm Humble: has quit [Ping timeout: 240 seconds] 17:51 < fromani> alitke: sorry I almost forgot your change in releaseVm 17:52 < fromani> alitke: in general we don't want to make the destroy path long 17:52 < alitke> Is it enough to just call setDaemon(True) on those threads 17:52 < alitke> ? 17:52 < alitke> then no need to wait() on them 17:52 < fromani> so my question is just... do we need wait for them? 17:53 < alitke> Well... I am concerned about an ongoing HSM operation and then the vm is started on another host 17:53 < fromani> alitke: fair point 17:53 * fromani looking deeper 17:55 < fromani> alitke: my feeling is this should be quite a rare case. Let's try to put things in perspective. How many of those threads we can have around? 17:56 < fsimonce> setDaemon is to make sigterm take place 17:56 < fsimonce> we trap sigterm iirc 17:56 < alitke> max is one per VmDisk 17:56 < fsimonce> and we may not go through vmRelease (or whatever is called) 17:56 < fromani> if setDaemon is ok for you guys it should be good for us as well 17:57 < fsimonce> it's pretty much a must if you want to be able to exit from the python process without waiting 17:58 < alitke> oh... And the sync is taking the resource locks on storage resources so it should be safe if the VM was restarted 17:58 < alitke> maybe? 17:58 < alitke> at least not likely to cause corruption 17:58 -!- dougsland (Douglas): has joined #vdsm 17:58 < fromani> fsimonce: well that's the "extrema ratio". We were hit hard in the (recent) past about releaseVm being too slow 17:58 < alitke> fsimonce, is that a fair statement re resource locks? 17:59 < fromani> on the other hand I don't expect a high number of LiveMergeCleanupThread around in the common case 18:00 < fsimonce> alitke, I didn't get the scenario 18:00 < alitke> let's say a cleanupThread is hung on a storage op and someone destroys a VM 18:00 < alitke> that cleanup thread would remain active 18:00 < fromani> yep that's the scenario I had in mind 18:00 < alitke> then the user starts the vm on another host 18:00 < fsimonce> it shouldn't, we're waiting right? 18:01 < alitke> we are talking about getting rid of the waits 18:01 < alitke> fromani, wants to keep releaseVM lean 18:01 < fromani> yep I'd love to make all the efforts on this direction. 18:01 -!- sbonazzo: has quit [Quit: Leaving.] 18:01 < fsimonce> no you can't get rid of those... we were talking about python waiting for all the threads to exit 18:01 < alitke> the issue at large is, can we trust the resource locks in this case or is the wait required? 18:02 < alitke> in the 99% case there will be 0 of these Threads 18:02 < fsimonce> if you don't set it as daemon, python is waiting for all the non-daemonic threads to exit 18:02 < alitke> and even if there are, the waits should be short in 99% of cases 18:02 < alitke> fsimonce, we'll keep setDaemon() 18:03 < alitke> but the wait() in releaseVM is a different issue 18:03 < alitke> vdsm is not exiting 18:03 < fsimonce> in 99% of the cases in which you have a live-merge completion on-going 18:03 < fsimonce> which is even rarer 18:04 < fsimonce> you need an ongoing live-merge completion, you need the storage to be unreachable... in that case you need to wait... but it's a really specific corner-case 18:04 < alitke> I feel safer keeping the waits in there 18:04 < fsimonce> yes indeed 18:05 < fsimonce> they'll really wait only in a really rare case 18:05 < fromani> well of course I trust you both if you say that it is a really rare case 18:05 < fromani> so please just add one short comment in the patch asserting that and I'm OK 18:05 < alitke> ok. Deal 18:06 -!- #vdsm dougsland: has quit [Read error: Connection reset by peer] 18:06 < fromani> ok thank you both 18:07 < alitke> sure, thank you 18:09 -!- dougsland (Douglas): has joined #vdsm 18:11 -!- #vdsm dougsland: has quit [Read error: Connection reset by peer] 18:12 -!- dougsland (Douglas): has joined #vdsm 18:18 -!- #vdsm dougsland: has quit [Read error: Connection reset by peer] 18:19 -!- dougsland (Douglas): has joined #vdsm 18:22 -!- #vdsm dougsland: has quit [Read error: Connection reset by peer] 18:23 -!- dougsland (Douglas): has joined #vdsm 18:26 -!- moolit1: has quit [Quit: Leaving.] 18:28 -!- #vdsm dougsland: has quit [Read error: Connection reset by peer] 18:30 -!- dougsland (Douglas): has joined #vdsm 18:34 -!- #vdsm dougsland: has quit [Read error: Connection reset by peer] 18:43 -!- dougsland (Douglas): has joined #vdsm 18:45 -!- #vdsm dougsland: has quit [Read error: Connection reset by peer] 19:15 -!- danken: has quit [Quit: Leaving.] 20:04 -!- phoracek: has quit [Ping timeout: 240 seconds] 20:12 -!- phoracek (phoracek): has joined #vdsm 20:22 -!- nsoffer: has quit [Read error: Connection reset by peer] 20:28 -!- apuimedo (antoni): has joined #vdsm 20:31 -!- nsoffer (Nir Soffer): has joined #vdsm 20:36 -!- apuimedo: has quit [Ping timeout: 250 seconds] 20:45 -!- phoracek: has quit [Quit: WeeChat 0.4.3] 20:51 -!- fabiand: has quit [Quit: Verlassend] 21:16 -!- bazulay: has quit [Ping timeout: 264 seconds] 21:19 -!- fromani: has quit [Remote host closed the connection] 21:47 -!- #vdsm ybronhei: has quit [Remote host closed the connection] 22:24 -!- bala: has quit [Quit: Leaving.] 22:33 -!- bala (purple): has joined #vdsm 23:03 -!- #vdsm dyasny: has quit [Ping timeout: 260 seconds] 23:11 -!- moolit (purple): has joined #vdsm 23:32 -!- bala: has quit [Quit: Leaving.] 23:50 -!- apahim: has quit [Quit: Leaving] 23:58 -!- bala (purple): has joined #vdsm --- Log closed vr jul 25 00:00:07 2014