18:16:15 #startmeeting networking_policy 18:16:16 Meeting started Thu Dec 15 18:16:15 2016 UTC and is due to finish in 60 minutes. The chair is SumitNaiksatam. Information about MeetBot at http://wiki.debian.org/MeetBot. 18:16:18 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 18:16:20 The meeting name has been set to 'networking_policy' 18:16:30 back to our favorite topic 18:16:38 #topic Remove expunge_all calls 18:16:43 #link https://review.openstack.org/#/c/399772/ 18:16:55 songole: hi 18:17:03 hi 18:17:09 songole: any chance that you can summarize the findings at your end? 18:17:44 SumitNaiksatam: I have asked Ashutosh to join. 18:17:55 songole: ah great 18:18:02 Don't see him online though 18:18:11 Too late for them in India 18:18:25 songole: completely understandable 18:18:44 based on his last email it seemed that he suspected that cascaded delete was not working as expected? 18:18:58 tbachman: you have thoughts on that? 18:19:17 SumitNaiksatam: I wasn’t sure. 18:19:18 He said something about avoiding things, but not fixing them 18:19:18 the gate passes tho 18:19:45 so, I wasn’t sure what else he was looking into 18:21:07 Are the tests passing with the fix he made? 18:21:20 the gate passes. No exceptions in the logs either 18:21:45 ok. thought he was seeing exceptions elsewhere 18:21:49 oh 18:21:57 I can go back and look at the latest 18:22:02 I just looked at the gate that was failing 18:22:50 tbachman: yeah the tests were passing 18:23:37 i am taking a look at the patch again 18:24:00 If I read it correctly, he thought that a cascade was causing something to be deleted, so when it later was going to be deleted explicitly, it failed because it was already gone 18:24:17 tbachman: ah okay 18:24:26 so now he is catching that exception 18:24:29 right 18:24:46 i guess we are good then :-) 18:24:49 but that was just from looking briefly at the change and his email 18:24:50 lol 18:24:56 right 18:25:04 yeah — it sounded like he felt there was more onion to peel. 18:25:05 i actually had not seen his most recent patch 18:25:06 just saw it 18:25:14 *patch set 18:25:19 (from the last email he sent) 18:25:23 maybe that’s changed 18:25:24 dunno 18:25:26 SumitNaiksatam: I was caught up with other things. didn't get a chance to sync up with him. 18:25:28 if this is all that is needed this looks good to me 18:25:34 songole: np 18:25:58 tbachman: how about we recheck this a few times today 18:26:01 I think the question is: if this is it, does this need to go hand-in-hand with the patch that removes the expunge? 18:26:04 SumitNaiksatam: sounds good 18:26:09 tbachman: i will keep a watch and do it 18:26:11 (i.e. two separate patches, or one) 18:26:17 SumitNaiksatam: thanks! 18:26:20 I can help there too 18:26:53 tbachman: i am fine either ways, since its the removal of expunge calls that reveals this, i am fine if the fix goes in the same patch 18:27:03 k 18:27:25 If this passes functional testing, then we can create the other patch that removes all of the expunge infrastructure 18:27:30 i think some explanation can be added as to how expunge_all was masking the issue 18:27:36 tbachman: sure 18:27:36 right 18:28:26 songole: thanks to you and your team for digging in to this 18:28:41 tbachman: in this patch we can put a comment in the place where we have commented out the calls to expunge_all as to why we are doing that and what is the effect of just having the yield in there (basically that its a no-op) 18:28:49 I’m not quite as familiar with this code, so it’s appreciated 18:28:49 songole: thanks indeed 18:28:58 SumitNaiksatam: sounds good 18:29:07 tbachman: SumitNaiksatam: I will pass it on... 18:29:15 tbachman: of course, BIG thanks to you for preservering with this investigation 18:29:22 SumitNaiksatam: np! 18:29:44 I just kicked things off :-) 18:30:04 (sound of ball dropping, and OneConvergence team picking it up) 18:30:14 lol 18:30:15 tbachman: lol 18:30:25 tbachman: you did way more than that 18:30:31 but this was a tough one 18:30:35 SumitNaiksatam: after this fix, can we remove the monkey patch in lb? 18:31:04 songole: you can try, but i dont it gaurantees 18:31:10 *dont think it 18:31:22 tbachman: thanks for all the help 18:31:34 since we were earlier discussing that ideally the db objects should be accessed within a transaction 18:31:34 songole: my pleasure — thank you and your team! 18:32:06 songole: i would recommend that we drop the lbaas patch and watch how it behaves 18:32:22 one… 18:32:22 SumitNaiksatam: ok. will do 18:32:23 (after we merge this patch) 18:32:24 step.... 18:32:25 at... 18:32:27 a… 18:32:34 time 18:32:36 lol 18:32:50 got the message :-) 18:33:04 tbachman: slowly 18:33:08 getting 18:33:09 it 18:33:09 lol 18:33:10 :) 18:33:12 :) 18:33:16 :-) 18:33:17 songole: nice 1 18:33:35 (sound of 9600 baud modem) 18:33:48 so, tbachman will you be adding the comments for the expunge_all call? 18:33:52 SumitNaiksatam: ack 18:33:53 tbachman: lol 18:34:04 I’ll add it to his patch 18:34:10 tbachman: right 18:34:10 (since he cherry-picked the expunge) 18:34:46 songole: once tbachman adds his comment, perhaps ashutosh can his investigation summary to the commit mesage (why the we catch objectdeleted exception, etc) and then we shoud be good 18:35:06 SumitNaiksatam: ok 18:35:08 *can add 18:35:10 SumitNaiksatam: will have that sometime after this meeting (am currently on VPN, and can’t push gerrits from my VM) 18:35:17 songole: thanks 18:35:47 tbachman: sure no hurry, we can let the current rebased run finish before we update the patch, so that we get the validation 18:35:55 SumitNaiksatam: thx 18:36:01 rkukura: plan sound okay to you? 18:36:21 sure 18:36:27 rkukura: okay, thanks 18:36:38 so we will wait until tomorrow to +2/A this 18:36:57 assuming (touch wood), this consistently passes the gate 18:37:25 #topic Open Discussion 18:37:46 FYI - the stable/liberty branches are being eol'ed 18:38:00 i have requested that the GBP branches be not eol'ed 18:38:05 and i dont think they will be touched 18:38:33 but since the other project branches will be eol’ed our devstack gate job will soon stop working on the stable/liberty patches 18:39:31 * tbachman just realized that he needs to check the EOL status of networking-cisco 18:39:36 so we will have to just rely on the py27 job once that happens (it hasnt happened yet, so the results of those jobs should still be taken into consideration when voting on a patch) 18:39:47 Same will happen for mitaka next, right? Do we know if the short cycle now will mean that is in less than 6 months? 18:39:55 tbachman: okay, i think there is a list 18:40:04 rkukura: :-( 18:40:07 SumitNaiksatam: thx 18:40:09 i wish you hadnt said that 18:40:12 lol 18:40:16 tbachman: i will check on the list 18:40:21 thx 18:40:28 SumitNaiksatam: don’t sweat it — I can dig it up 18:40:49 SumitNaiksatam: rkukura just gave you more to work on anyway ;) 18:40:59 tbachman: :-( 18:41:01 lol 18:41:28 our rally job has been broken for a few days 18:41:59 it was actually broken for other projects too, i have to check how they fixed it 18:42:16 we might have to cap some setuptools dependency 18:42:38 other that i am gettting the aim job to work now 18:42:42 hopefully it should be ready soon 18:43:35 meeting logistics - there are no openstack project meetings happening over the next couple of weeks 18:44:04 we wil meet next week only if there is some urgent need, otherwise the meeting is cancelled for the next couple of weeks 18:44:14 if we are meeting next week, i will ping you, else assume we are not 18:44:20 tbachman: i know you will be on PTO 18:44:30 SumitNaiksatam: ack. From 20th - 27th 18:44:37 tbachman: so dont bother about this even if we end up meeting 18:44:44 k 18:44:48 i will be on leave too 18:44:56 anything else that we need to discuss? 18:45:22 its a big relief that we are finally making progress on the expunge_all stuff! 18:45:22 not that I can think of 18:45:26 rkukura: ok thanks 18:45:49 songole: any parting year end thoughts? 18:46:07 “parting” could also be interpreted as “partying” ;-) 18:46:21 SumitNaiksatam: happy holidays .. :) 18:46:27 songole: nice one :-) 18:46:53 tbachman: thanks again for all your work, anything else? 18:47:13 SumitNaiksatam: not from me, other than to wish all a happy holidays! 18:47:32 yeah, thanks for all the work during the year, and happy holidays to all@ 18:47:35 ! 18:47:40 thanks for joining today 18:47:43 bye all! 18:47:46 bye 18:47:47 SumitNaiksatam: thank you! 18:47:51 bye! 18:47:52 #endmeeting