16:00:06 <klindgren> #startmeeting Performance Team
16:00:11 <openstack> Meeting started Tue Mar 29 16:00:06 2016 UTC and is due to finish in 60 minutes.  The chair is klindgren. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:15 <openstack> The meeting name has been set to 'performance_team'
16:00:17 <klindgren> awesome
16:00:59 <klindgren> SO looking back at last meeting.  Seems like we had some action items from the last one
16:01:35 <klindgren> #topic Action Items
16:01:43 <lezbar> Hi, I'm new here
16:02:17 <klindgren> Hi
16:02:34 <klindgren> so looks like something about osprofiler-engines:
16:02:37 <klindgren> #link https://etherpad.openstack.org/p/osprofiler-engines
16:03:01 <rohanion> \o guys
16:03:24 <klindgren> anyone have any updates on that?
16:04:09 <rohanion> I don't have any updates regarding osprofiler engines, I've spent my week playing around with the insides of osprofiler
16:04:16 <klindgren> kk
16:04:28 <klindgren> moving on then
16:04:31 <klindgren> #link https://review.openstack.org/#/c/291657/
16:04:53 <klindgren> Looks like that got merged a few days ago
16:05:08 <klindgren> which is adding a new porject of performa
16:05:29 <klindgren> ilyashakhat - nice work on that
16:06:34 <bsilverman> How does performa differ from Rally? In general?
16:07:05 <boris-42> bsilverman: I believe it should be just rally runner
16:07:13 <bsilverman> ok, thanks.
16:07:30 <boris-42> bsilverman: however we have some work to do in Rally
16:07:39 <boris-42> bsilverman: to get scalability
16:08:26 <klindgren> did anyone else have an action items besides dina?
16:08:33 <klindgren> #link http://eavesdrop.openstack.org/meetings/performance_team/2016/performance_team.2016-03-22-16.00.html
16:08:51 <klindgren> those seem to be the only action items from last meeting
16:09:51 <listomin> I can share my status with testing provisioning systems
16:10:07 <klindgren> Sure
16:10:17 <klindgren> #topic testing provisioning systems update
16:10:25 <listomin> I've tested two installations of Cobbler provisioning system:
16:10:34 <listomin> When a node downloads kernel and initrd via http protocol and via tftp
16:10:40 <listomin> I mean the step after pxelinux loaded
16:11:19 <listomin> So, with http variant I could provision 100% nodes with only number of nodes less then 40
16:11:45 <listomin> I'll make a commit with the results soon
16:11:56 <listomin> tftp way works fine
16:12:20 <klindgren> Interesting - so tftp was fine for nodes greater than 40?
16:12:31 <listomin> yep. up to 195
16:12:59 <listomin> all from my side
16:13:11 <klindgren> Outside of the failure rate - when doing a small number of nodes was one mode faster than the other?
16:14:03 <klindgren> In playing with ironic I noticed that tfpting the ipa image via pxelinux was a minute or two slower than tftp an ipxe boot rom and having the ipxe bootrom download the IPA image/ramdisk via http
16:14:04 <listomin> It seems tftp work faster, but I didn't collect such results
16:14:19 <klindgren> kk
16:14:40 <listomin> one more
16:15:00 <listomin> During the investigation of the failures I've found
16:15:25 <listomin> That Apache sends not all file, just part of the linux kernel
16:15:53 <listomin> Then Apache waits 300 sec (timeout) and answer 200 http code
16:16:16 <listomin> It seems some issue on the pxe http client side
16:16:38 <listomin> now all from my side :)
16:16:43 <klindgren> :-)
16:17:55 <klindgren> Is anyone here that can talk about the test plans statuses?
16:18:21 <klindgren> listomin, one patchesset looks like it was from you
16:18:26 <klindgren> looks like both are merged
16:18:34 <klindgren> #link https://review.openstack.org/#/c/295308/
16:19:04 <klindgren> the other seems to have been ilyashakhat
16:19:12 <klindgren> #link https://review.openstack.org/#/c/295919/
16:19:35 <listomin> yep, it was merged already
16:19:42 <klindgren> cool
16:20:00 <klindgren> #topic test plan statuses
16:20:06 <klindgren> all merged!
16:20:16 <klindgren> #topic OSProfiler weekly Update
16:20:37 <klindgren> Ayone have any update to share on OSProfiler?
16:21:41 <klindgren> It looks like all of these are DinaBelova
16:23:16 <klindgren> ilyashakhat, rohanion what are your updates?
16:23:28 <klindgren> since DinaBelova said that you had some before the meeting started
16:24:40 <rohanion> well, I want to share with boris-42 and DinaBelova a mockup of a new osprofler architecture
16:25:02 <boris-42> rohanion: ?
16:25:21 <rohanion> oh hey
16:25:42 <rohanion> will you be available for a call in an hour or so?
16:27:53 <rohanion> the summary is: I want to get rid of Notifier class completely, and I think that the Driver instance should be a field of the Profiler class
16:30:06 <klindgren> So I assume that you will be following up with boris-42 and DinaBelova with a call to go over the purposed changes?
16:30:13 <rohanion> and I think about the possibility of `osprofiler trace show` command to work with a Profiler instance instead of parsing ceilo directly
16:30:19 <rohanion> yes
16:30:55 <boris-42> rohanion: can we postpone call to tomorrow
16:31:03 <rohanion> np
16:31:05 <rohanion> sure
16:31:07 <boris-42> rohanion: great
16:31:25 <klindgren> alrighty
16:31:50 <klindgren> #topic Open Discussion
16:32:33 <klindgren> Anyone have any open things to discuss?
16:34:47 <klindgren> hearing nothing - I am going to end this meeting
16:35:09 <harlowja> end it! :-P
16:35:10 <klindgren> #endmeeting