15:00:24 <yuval> #startmeeting karbor 15:00:25 <openstack> Meeting started Tue Mar 21 15:00:24 2017 UTC and is due to finish in 60 minutes. The chair is yuval. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:26 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:28 <openstack> The meeting name has been set to 'karbor' 15:00:31 <yuval> Hello everybody 15:00:35 <yuval> and welcome to the weekly karbor meeting 15:00:44 <zhonghua> hi 15:00:51 <chenying> hi 15:01:23 <yuval> #info zhonghua and chenying in meeting 15:01:53 <yuval> waiting for anyone else? zhangshuai? 15:02:23 <yuval> let's begin 15:02:26 <zhonghua> sorry, I donot know 15:02:35 <yuval> #topic fullstack 15:02:44 <yuval> chenying: you added this item 15:02:50 <chenying> yes 15:03:22 <chenying> I want to add a new fullstack job, focus on the operationengine tests. 15:03:32 <chenying> What's your oppion about it? 15:03:43 <yuval> chenying: add new fullstack jobs? sounds great 15:03:56 <yuval> chenying: wait, new job or new tests? 15:04:13 <chenying> IMO, the fullstack takes too much time. 15:04:32 <chenying> new job, so we can move some fullstack tests to new job. 15:05:30 <yuval> chenying: do you know how much time each test takes right now? 15:05:45 <yuval> chenying: why not enable concurrency, as zhangshuai is trying to do? 15:06:42 <chenying> As you know, zhangshuai works on this patch for a long time. 'Update fullstack efficiency' There are some error about swit when parallelly runing tests. 15:07:17 <yuval> chenying: ok, and? 15:08:47 <chenying> The reason is that, the swif will handle the request parallelly, the time is too long and the rcp will timeout, because of the swift runing the vm, the vm don't have high performance. 15:08:58 <yuval> chenying: if zhangshuai ran into a swift error, there might be real issue there, that we need to solve. Separating the tests into two gate jobs will not resolve that 15:09:13 <yuval> chenying: are you sure this is the issue? 15:09:45 <chenying> yuval: yes I and zhangshuai test this patch for one or two days. 15:10:01 <yuval> chenying: how do you know the problem is swift performance? 15:10:33 <chenying> yuval: When cinder call swift api to do volume backup, it may take 4 or 5min to backup a volume. 15:10:51 <yuval> chenying: wait 15:11:09 <yuval> chenying: the main issue about zhangshuai patch is, that swift _fails_ for some calls 15:11:10 <chenying> yuval in the situation runing the test parallelly. 15:11:16 <yuval> chenying: not the long time it takes 15:11:36 <yuval> chenying: besides, how do you know swift is the bottleneck? 15:12:16 <chenying> yuval: No The reason is taking too long . 15:12:31 <yuval> chenying: I do not understna 15:12:33 <yuval> *understand 15:12:49 <yuval> chenying: please reiterate 15:13:25 <chenying> yuval: The reason is that swift taking too long time, and the rpc will timeout. 15:15:27 <yuval> chenying: do you have something to support that claim? log file? 15:17:17 <chenying> Yes log file, We have calculated the handling time of swift by the time log of rpc messages id when cinder call the backup api(backup colume to swift). 15:17:59 <chenying> In the situation runing the test parallelly It take about 4 -5 min. 15:18:57 <yuval> chenying: very interesting. Can you refer me to the log messages later? Maybe we can do something to resolve that in the same gate job. If we don't have a choice, we will split into two different gate jobs 15:21:01 <chenying> zhangshuai can provide the log file about using swift to backup volume tomorrow. I don't have it now. 15:21:13 <yuval> chenying: great 15:21:28 <yuval> chenying: great job 15:22:38 <yuval> #topic pike 15:22:58 <chenying> https://etherpad.openstack.org/p/karbor-Pike-planning 15:23:00 <yuval> #link https://ethercalc.openstack.org/karbor-pike 15:23:23 <yuval> chenying: never updated that etherpad 15:23:27 <yuval> chenying: see the ethercalc 15:23:38 <chenying> yuval We have added our name to the task that want to do. 15:23:58 <yuval> also, please do not change or add tasks, just assign to yourself 15:24:22 <yuval> it is sorted by priorirty and size (estimation) 15:25:39 <chenying> OK I will ask other guys to update this link tomorrow. 15:25:46 <yuval> also, we have a launchpad milestone: 15:25:49 <yuval> #link https://launchpad.net/karbor/+milestone/pike-ms 15:26:51 <yuval> chenying: again, please update only the assignee 15:26:57 <yuval> (carrier) 15:27:53 <chenying> I know. 15:29:32 <yuval> #topic Open Discussion 15:31:01 <yuval> Anything else? 15:31:32 <chenying> I don't have any topics. 15:31:39 <yuval> Have a good night! 15:31:41 <yuval> :) 15:31:43 <yuval> #endmeeting