--- Log opened vr jan 09 00:00:28 2015 00:28 -!- fsimonce: has quit [Quit: Coyote finally caught me] 01:03 -!- mtayer (purple): has joined #vdsm 01:26 -!- apahim: has quit [Ping timeout: 264 seconds] 01:31 -!- mtayer: has quit [Quit: Leaving.] 01:32 -!- adahms (Andrew Dahms): has joined #vdsm 01:42 -!- jbrooks: has quit [Ping timeout: 245 seconds] 02:16 -!- #vdsm ybronhei: has quit [Quit: Leaving.] 02:16 -!- ybronhei (purple): has joined #vdsm 03:35 -!- jbrooks: has quit [Ping timeout: 265 seconds] 03:42 -!- bala: has quit [Quit: Leaving.] 04:28 -!- #vdsm dyasny: has quit [Ping timeout: 265 seconds] 05:39 -!- Madhu_ (Madhu_): has joined #vdsm 07:09 -!- tomoe (tomoe): has joined #vdsm 07:29 -!- timothy (Timothy Asir): has joined #vdsm 07:29 -!- timothy is now known as Guest13209 08:02 -!- sbonazzo (purple): has joined #vdsm 08:14 -!- apuimedo (antoni): has joined #vdsm 08:26 -!- pkliczew (Piotr Kliczewski): has joined #vdsm 08:30 -!- adahms: has quit [Quit: Leaving] 08:39 -!- fromani (Francesco Romani): has joined #vdsm 08:41 -!- mskrivanek_away is now known as mskrivanek 09:08 -!- nsoffer (Nir Soffer): has joined #vdsm 09:27 -!- tomoe: has quit [Remote host closed the connection] 09:42 -!- fsimonce (Federico): has joined #vdsm 09:46 -!- fabiand (Fabian Deutsch): has joined #vdsm 09:50 -!- georgeh_: has quit [Ping timeout: 245 seconds] 10:00 -!- nsoffer: has quit [Ping timeout: 244 seconds] 10:10 -!- georgeh (George Hicks): has joined #vdsm 10:35 -!- bala: has quit [Ping timeout: 264 seconds] 11:23 < fromani> danken: good morning, here's a couple of silly harmless patches which could go in anytime: http://gerrit.ovirt.org/#/c/36656/ and http://gerrit.ovirt.org/#/c/36717/ 11:23 * fromani is fixing bigger and more meaningful patches 11:23 < apuimedo> danken: licensing fixed ;-) 11:23 < apuimedo> Thanks for pointing it out 11:25 -!- #vdsm ybronhei: has quit [Quit: Leaving.] 11:25 -!- ybronhei (purple): has joined #vdsm 11:35 <@danken> apuimedo: I could have been a lawyer ;-) 11:35 <@danken> apuimedo: (I did not glimpse on anything else there, though) 11:40 < apuimedo> danken: it should be pretty simple, only getting an environment for it is complicated 11:40 < apuimedo> :P 11:40 < apuimedo> building the engine, etc 11:46 < apuimedo> danken: what is remaining to get vdsm to be noarch instead of x86_64? 11:47 -!- bala: has quit [Ping timeout: 255 seconds] 11:54 < fromani> safelease, I guess 12:06 < apuimedo> fromani: right 12:07 < fromani> apuimedo: IIRC there is an ongoing effort to spawn it off VDSM main tree 12:08 < apuimedo> fromani: who's on it? 12:08 < fromani> apuimedo: IIRC vdelima 12:08 < apuimedo> aha 12:08 < apuimedo> well, it could also easily be made with python+cffi/ctypes 12:09 < fromani> not sure, isn't safelease a tool? like udevadm, iscsiadm and stuff 12:11 < apuimedo> fromani: it is 12:11 < apuimedo> but you can also make tools in python ;-) 12:11 < fromani> apuimedo: sure, I just don't think cffi/ctypes could be of help for this task 12:12 < fromani> but I just had a quick glance to safelease back in time 12:32 <@danken> safelease is an ancient legacy. I don't want to re-implement it. just to throw it out of vdsm 12:33 < apuimedo> ok ;-) 12:39 <@danken> fromani: giving bazulay's objections, you can drop the Bug-Url from http://gerrit.ovirt.org/#/c/36626/2/vdsm/numaUtils.py so as not to claim to solve the problem. and change the side issue to the main subject of the commit message: 12:39 <@danken> replace slow minidom in yet another spot. 12:39 < fromani> danken: sure thing 12:40 < fromani> just a few minutes, I'm in the middle of a complex rebase 12:41 <@danken> fromani: another thing could be to move this function to the sampling thread, and make get() pull it from the cache. it does not save CPU, but improves latency. 12:42 < fromani> danken: I had a long and painful time wrapping my head around that code (once again, my bad to have it acked so quickly), I just need a few more context to be able to wipe it away alltogether (of course without losing features) 12:43 <@danken> fromani: take your time 12:43 <@danken> fromani: I'm not sure that it is such a horrible regression as bazulay claims it to be. 12:44 < fromani> danken: we had more developments, turns out numaUtils code is "just" ugly, but not the culprit for horrible performance 12:44 -!- tomoe (tomoe): has joined #vdsm 12:44 < fromani> emarciano just commented out the line that kicked in that code, with no performance gains 12:49 <@danken> fromani: that was my gut feeling... but we still have the question regarding what DOES eat the time 12:50 < fromani> danken: I have vague feelings, not hard data to back them up. 12:50 <@danken> fromani: and what's your guess? 12:52 -!- tomoe: has quit [] 12:52 < fromani> danken: biggest problem is: I don't understand why performance wasn't so horrible in the previous scale tests. That said, I believe that we "simply" have too many threads _fighting for CPU_. Threading in cpython is still bearable as long as they do I/O. Otherwise performance becomes horrible. 12:53 < fromani> danken: OTOH, it could be something peculiar on that host, but I have no clue yet. 13:13 -!- Guest13209: has quit [Remote host closed the connection] 13:21 -!- bala: has quit [Remote host closed the connection] 13:34 -!- Madhu_: has quit [Quit: Madhu_] 13:41 < fromani> danken: done about http://gerrit.ovirt.org/#/c/36626/ 13:42 < fromani> danken: I collected a few hopefully-harmless-being-mostly-renaming cleanups in the patches following 36626 14:04 <@danken> fsimonce: fromani: I know that I'm nagging allot about it, but please approve http://gerrit.ovirt.org/#/c/36434/ 14:04 <@danken> I hate this broken test that we have 14:08 -!- sbonazzo: has quit [Quit: Leaving.] 14:10 -!- sbonazzo (purple): has joined #vdsm 14:10 < fromani> danken: np, let me just dust off my script-shell-fu to properly followup this patch 14:14 -!- danken: has quit [Quit: Leaving.] 14:14 -!- mode/#vdsm: by ChanServ 14:14 -!- danken (purple): has joined #vdsm 14:39 -!- bala: has quit [Ping timeout: 244 seconds] 15:00 -!- dyasny (Dan Yasny): has joined #vdsm 16:18 -!- mskrivanek is now known as mskrivanek_away 16:53 <@danken> apahim: going back to http://gerrit.ovirt.org/#/c/31529/25/vdsm/storage/iscsi.py we're trying to see if there's a nice way to simplify the code 16:56 <@danken> fsimonce is right about the code being so repetitious... 17:00 <@danken> but I don't have a very good suggestion in mind 17:01 <@danken> apahim: clearly the name "needsLooseRpFilter" is misleading 17:03 <@danken> apahim: if we merge the two functions setLooseRpFilterIfNeeded/restoreStrictRpFilterIfNeeded into one, we can get rid of needsLooseRpFilter by inlining it. 17:04 -!- mskrivanek_away is now known as mskrivanek 17:04 <@danken> how about setRpFilterIfNeeded(netIfaceName, hostname, loose_or_strict_boolean) ? 17:05 -!- pkliczew: has quit [Ping timeout: 256 seconds] 17:06 -!- apuimedo: has quit [Ping timeout: 244 seconds] 17:06 < apahim> ok danken... I will work on this right now. 17:07 <@danken> apahim: sec, I'd love to hear fsimonce's opion on this 17:07 < apahim> ok :) 17:07 <@danken> after all, he broght the issue up 17:08 < fsimonce> setRpFilterIfNeeded(netIfaceName, hostname, loose_or_strict_boolean) sounds good... it will remove the duplicated code 17:09 < fsimonce> also remember that in the strict part you can probably just check len(session) because the session was just removed 17:11 < apahim> ok fsimonce 17:13 <@danken> apahim: I hope we can inline _needsLooseRpFilter, but if not, _needsRpFilterMongering is a less misleading name 17:14 -!- sbonazzo: has quit [Quit: Leaving.] 17:46 -!- mskrivanek is now known as mskrivanek_away 18:04 -!- bala: has quit [Quit: Leaving.] 18:31 -!- evilissimo1 (Vinzenz Feenstra): has joined #vdsm 18:31 -!- msivak_ (Martin Sivák): has joined #vdsm 18:31 -!- mskrivanek (mskrivan): has joined #vdsm 18:33 -!- evilissimo: has quit [Read error: Connection reset by peer] 18:33 -!- fabiand: has quit [Quit: Verlassend] 18:35 -!- msivak: has quit [Ping timeout: 255 seconds] 18:35 -!- mskrivanek_away: has quit [Ping timeout: 265 seconds] 18:36 -!- evilissimo1: has quit [Ping timeout: 245 seconds] 18:37 -!- mskrivanek: has quit [Ping timeout: 264 seconds] 18:37 -!- msivak_: has quit [Ping timeout: 264 seconds] 18:38 -!- mskrivanek (mskrivan): has joined #vdsm 18:38 -!- msivak_ (Martin Sivák): has joined #vdsm 18:39 < apahim> danken, fsimonce http://gerrit.ovirt.org/#/c/31529/26/vdsm/storage/iscsi.py 18:50 -!- evilissimo1 (Vinzenz Feenstra): has joined #vdsm 19:00 <@danken> thanks, have a nice weekend 19:58 < apahim> fsimonce, I fixed the log position. thank you for the review. 20:04 -!- vered (Vered Volansky): has joined #vdsm 20:15 -!- #vdsm dyasny: has quit [Ping timeout: 264 seconds] 20:33 -!- danken: has quit [Quit: Leaving.] 21:43 -!- apahim: has quit [Quit: Leaving] 22:07 -!- vered: has quit [Ping timeout: 264 seconds] 23:05 -!- fromani: has quit [Quit: Leaving] --- Log closed za jan 10 00:00:29 2015