08:00:37 #startmeeting storlets 08:00:38 Meeting started Tue Dec 6 08:00:37 2016 UTC and is due to finish in 60 minutes. The chair is eranrom. Information about MeetBot at http://wiki.debian.org/MeetBot. 08:00:39 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 08:00:41 The meeting name has been set to 'storlets' 08:00:46 Hi 08:01:44 Hi 08:01:51 akihito: Hi 08:02:09 hello 08:02:13 kota_: Hi 08:02:18 hi 08:02:58 Do you know if Takashi plans on joining? 08:03:31 he was willing to join here 08:03:58 he said when I asked 2 hours ago 08:04:18 eranrom:^^ 08:04:28 He is absent... 08:04:30 ok, We need him for the PTG topic 08:04:50 and just get his response, it looks like he got another troublesome 08:05:05 I see. Hope he is ok. 08:05:35 So lets start with the other topic: 08:05:52 so please go ahead today's topics 08:06:10 the other topic? 08:06:11 #topic Big-Tent 08:06:14 k 08:06:47 Any comments on the steps I have suggested (https://wiki.openstack.org/wiki/Meetings/Storlets#Agenda:) 08:06:59 list is on item 3 08:07:22 looking 08:07:50 yeah, looks great for the agenda 08:08:00 and now on item3 ok 08:08:48 eranrom: do you think of we need to merge all of them? 08:09:24 eranrom: or hope landing part of them and propose to get offical in parallel? 08:09:35 I guess the latter two are not really required 08:10:52 I can get back to Flavio after we merge the first two 08:10:57 and continue from there. 08:11:00 sounds good idea, we would include them as in our load map but not commitments 08:11:06 k 08:11:12 ok. 08:11:32 #agreed 08:11:35 on the v3 patch, probably I have to do that in my homework. 08:11:58 because it looks like takashi has added +2 already, right? 08:11:59 kota_: sure 08:12:03 yep 08:12:19 let me check the patch link.. 08:12:39 #link https://review.openstack.org/#/c/394168/ 08:12:43 this one? 08:12:57 yes 08:13:18 * eranrom will take care of headers while kota_ does homework :-) 08:13:43 er, recently, devstack gate has been changed keystone v3 default, we need that due to using devstack, make sense. 08:13:59 eranrom: thanks for taking care of headers! 08:14:31 kota_: sure. yes we probably need that. Takashi wants it for quite a while now. 08:14:36 eranrom: quick question 08:14:57 eranrom: why do we need to update keystone v3 in *ansible scripts*? 08:15:38 eranrom: is it not used for building testing environment anymore? 08:16:07 I'm just looking at file list modified 08:16:13 The ansible scripts perform various Swift operations that require a token, and so they need to authenticate using v3 now... 08:17:11 Specifically, there is a role that enables an account for storlets. This requires creating several containers, and setting the storlet enabled attribute on the account 08:17:36 er... 08:17:49 k, getting, getting 08:18:16 :-) 08:18:29 so the devstack is used just as instration for the gate now and we still need to setup something (e.g. account) via ansible 08:18:52 and then, current script uses v2-like so that we need to update them. 08:19:00 right? 08:19:42 s/instaration/installation/ 08:19:45 kota_: yes. The interesting point for me was to discover that although devstack installed Keystone with V3 API the current code was still working with V2 API 08:19:50 For me it was a surprise. 08:20:15 eranrom: interesting 08:20:31 keystone is full of surprises :-) 08:20:37 eranrom: and thanks, that info is *SUPER* helpful for my homework :P 08:20:56 kota_: sure 08:21:05 This is all I had, so moving to open discussion 08:21:12 #topic Open Discussion 08:21:58 Anything else? 08:22:15 eranrom: not a big progress right now 08:23:01 kota_: Saw you posted a long running test patch. I guess you are looking at a long running storlet at the proxy. 08:23:49 not specifically only on the proxy but for now I'm making sure which param should work well on the timeout cases 08:24:13 to describe my co-worker who dosn't know the swift/storlets config in detail :/ 08:24:57 since that, he was starting to increase all timeout value up to enough to run the heavy task 08:25:32 but i don't want to do that because w/o knowledge increasing timeout may cause side-effect! 08:26:31 so that, now I'm making sure if storlet_timeout can work as I expected and we could use also node timeout for swift conf in default value. 08:26:40 even if it's heavy task. 08:26:42 kota_: right. Note that when running on an object node you may get different timeouts then when running on the proxy. 08:26:44 it's a 08:26:56 eranrom: exactly 08:27:09 kota_: different code path... 08:27:18 painful :/ 08:27:38 kota_: indeed. Doron and I have investigated this a lot in the past. 08:28:00 eranrom: in older days? 08:28:09 kota_: yes, abut a year ago 08:28:46 We have mapped the problem, but we did not come up with a solution. 08:28:46 eranrom: i think, with recent work, the refactoring we have done would make a good shape to abstract the path for timeout 08:28:57 oh, maybe? 08:29:04 that problem still be there? 08:29:24 kota_: well, if you do not want to play with node timeout then yes, it is still there. 08:29:41 and may involve swift wsgi and/or eventlet code 08:29:46 eranrom: if you could comment to point out the problem in my patch, it could help me to understand. 08:30:20 kota_: Will do, I will recollect our findings first :-) 08:30:28 eranrom: thanks! 08:30:43 kota_: my pleasure! 08:31:54 that's all from me today. 08:32:04 kota_: thanks 08:32:25 If there is nothing else we will get back 30 mins :-) 08:32:46 :-) 08:33:18 that's good because I can use the 30 mins for the review of keystone v3 patch 08:33:23 :-) 08:33:28 ok so in that case: 08:33:34 Thank you all for joining. 08:33:46 See you on #openstack-storlets 08:33:53 #endmeeting