| *** _ducttape_ has quit IRC | 00:00 | |
| rm_work | lol k | 00:01 |
|---|---|---|
| rm_work | i broke it | 00:01 |
| rm_work | well, give me a sec on rules | 00:01 |
| johnsom | I just checked out master to double check | 00:01 |
| rm_work | almost got the admin_state thing done | 00:01 |
| rm_work | oh | 00:01 |
| rm_work | you mean v1 isn't working on master? | 00:02 |
| rm_work | i thought you said it was T_T | 00:02 |
| johnsom | No, I'm going to double check that it does work on master.... | 00:02 |
| rm_work | ah k | 00:02 |
| rm_work | yeah i am pretty sure i just broke it | 00:02 |
| rm_work | it worked with sqlite but <_< | 00:03 |
| rm_work | sometimes things are wonky | 00:03 |
| johnsom | No, ok, it doesn't work for me on master v1 either | 00:04 |
| rm_work | ah | 00:04 |
| rm_work | >_< | 00:04 |
| rm_work | well i'm going to revert the changes I made around trying to fix it | 00:04 |
| rm_work | and then we can file a bug | 00:04 |
| *** ducttape_ has joined #openstack-lbaas | 00:04 | |
| johnsom | Ok | 00:05 |
| johnsom | I wonder if it is this screwy sqlalchemy stuff. If I say position 6 does this reorder junk think it's really 1 because it's the only policy????? | 00:06 |
| johnsom | That would be confusing as heck for users as it prints out this position # | 00:06 |
| rm_work | yeah | 00:08 |
| rm_work | if you put in position 1000 it will reorder it to 1 | 00:08 |
| rm_work | that was by design | 00:09 |
| johnsom | If a new policy is created with a position that matches that of an existing policy, then the new policy is inserted at the given position. | 00:09 |
| johnsom | If a new policy is created without specifying a position, or specifying a position that is greater than the number of policies already in the list, the new policy will just be appended to the list. | 00:09 |
| johnsom | Per stephen's docs | 00:09 |
| johnsom | So, ok, maybe pebkac | 00:09 |
| johnsom | We probably should have called that rank in the list/show and position in the set | 00:10 |
| rm_work | pushing up the admin_state fix in one sec, along with a rollback to my changes to the policy reorder code, so i don't confuse things when someone tries to go fix it | 00:10 |
| rm_work | i see some cleanup that could be done | 00:11 |
| rm_work | but it'll take a deeper look | 00:11 |
| johnsom | Ok, I need to sign off pretty soon for the evening | 00:11 |
| rm_work | hmm k | 00:12 |
| rm_work | same, but | 00:12 |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: Align Octavia API to n-lbaasv2 for L7Rules https://review.openstack.org/406336 | 00:12 |
| openstackgerrit | Adam Harwell proposed openstack/octavia master: Align Octavia API to n-lbaasv2 for L7Policy https://review.openstack.org/406328 | 00:12 |
| rm_work | there you go :P | 00:12 |
| johnsom | Ok, if admin is good and we rolled back the position stufff, this is probably good to go | 00:12 |
| rm_work | need you to test it | 00:13 |
| rm_work | but it should be | 00:13 |
| rm_work | hopefully a quick test since i'm not messing with migrations anymore | 00:15 |
| johnsom | Yeah, working on it | 00:15 |
| rm_work | can you pastebin your test commands? | 00:17 |
| rm_work | i will try to track down this policy ordering thing | 00:17 |
| rm_work | have the debugs in place | 00:17 |
| rm_work | yeah ok I'll see if i can track down this policy reorder issue... maybe later today | 00:20 |
| rm_work | gonna head out for now though | 00:21 |
| *** ducttape_ has quit IRC | 00:22 | |
| johnsom | I think it works and I was just using it wrong.... | 00:22 |
| johnsom | These pass my tests now, so +2s from me | 00:22 |
| rm_work | oh | 00:24 |
| rm_work | erp | 00:24 |
| rm_work | reorder is for like | 00:24 |
| rm_work | create two policies, A(position=1) and B(position=2) | 00:25 |
| rm_work | then update B(position=1) | 00:25 |
| rm_work | and it should make B=1 A=2 | 00:25 |
| rm_work | when you get A and B | 00:25 |
| rm_work | but, sweet! | 00:25 |
| rm_work | diltram / xgerman: your court now :) | 00:25 |
| eandersson | Have anyone seen very long response times to get_loadbalancers? We don't have a crazy amount of lbs, but seeing super slow api requests (50s+). | 00:28 |
| johnsom | Yeah, reorder seems to work. | 00:28 |
| rm_work | lol ok | 00:28 |
| rm_work | well, glad i backed out my changes | 00:28 |
| johnsom | eandersson No, it should be reasonably quick, it is just a database query. | 00:29 |
| eandersson | Yea - we are seeing the db query take 50s | 00:29 |
| eandersson | or more | 00:29 |
| johnsom | Ouch, that is not normal | 00:30 |
| johnsom | That was with the mysql client? (assuming you are using mysql) | 00:30 |
| eandersson | yep mysql (pymysql) | 00:30 |
| johnsom | Can you paste the query you made? The select statement? | 00:31 |
| eandersson | sorry - we haven't been able to perform the mysql query manually using the client yet | 00:31 |
| johnsom | Oh, ok, that was what I was asking. | 00:31 |
| eandersson | https://github.com/openstack/neutron-lbaas/blob/master/neutron_lbaas/db/loadbalancer/loadbalancer_dbv2.py#L377 | 00:32 |
| eandersson | That function takes almost 60s to process. | 00:32 |
| eandersson | We have like 200 lbs in our db | 00:33 |
| *** ducttape_ has joined #openstack-lbaas | 00:34 | |
| johnsom | Yeah, I have done tests with > 2000 and haven't had problems like that | 00:34 |
| *** yuanying has joined #openstack-lbaas | 00:34 | |
| johnsom | Do you have a lot of churn? A lot of create/delete cycles? | 00:35 |
| eandersson | I wouldn't say a crazy amount. | 00:35 |
| johnsom | If so, there are mysql maintenance tasks the mysql folks recommend you run. | 00:35 |
| johnsom | This kind of thing: | 00:36 |
| johnsom | https://dev.mysql.com/doc/refman/5.7/en/mysqlcheck.html | 00:36 |
| johnsom | But, be aware some of those command will lock the tables, so could slow/block other activities like new creates | 00:37 |
| *** asdf__ has joined #openstack-lbaas | 00:38 | |
| johnsom | slow disks and low memory can also impact mysql. Check and make sure the array with the DB in it isn't running in degraded mode.... | 00:39 |
| eandersson | Everything else works fine though | 00:39 |
| eandersson | Tried a second site now, much slower, almost no activity in terms of lb (got maybe 20lbs) and takes about 15s | 00:39 |
| johnsom | Ok. Yeah, I would look at the table maintenance tasks. Probably first I would do a manual select on the DB for that table. | 00:40 |
| johnsom | I have to run, but maybe tomorrow morning I will write a little script and load up a few 100 lbs and give it a go. | 00:43 |
| rm_work | still shouldn't take 15s for 20lbs | 00:45 |
| rm_work | should take like 1s for 1000lbs <_< | 00:45 |
| johnsom | Yep | 00:45 |
| eandersson | So we are unable to find a slow mysql query, it is most likely getting stuck building the result from the query. | 00:46 |
| johnsom | So a direct select on the table returned as you would expect? | 00:47 |
| eandersson | We turned on slow query logs and nothing longer than a second. | 00:53 |
| eandersson | When you ran tests with 2k lbs, did you have a lot of members as well? | 00:53 |
| reedip | o/ | 01:14 |
| *** reedip has quit IRC | 01:20 | |
| *** amotoki has joined #openstack-lbaas | 01:31 | |
| *** ducttape_ has quit IRC | 01:37 | |
| *** ducttape_ has joined #openstack-lbaas | 01:46 | |
| *** ducttape_ has quit IRC | 02:07 | |
| *** ducttape_ has joined #openstack-lbaas | 02:14 | |
| openstackgerrit | Alexander Amerine proposed openstack/octavia master: blueprint LBaaS Alternative Monitoring IP/Port https://review.openstack.org/453001 | 02:22 |
| *** gongysh has joined #openstack-lbaas | 02:27 | |
| *** reedip has joined #openstack-lbaas | 02:44 | |
| *** armax has quit IRC | 03:13 | |
| *** links has joined #openstack-lbaas | 03:16 | |
| openstackgerrit | Michael Johnson proposed openstack/octavia master: Adds v2 load balancer API section https://review.openstack.org/438757 | 03:29 |
| *** ducttape_ has quit IRC | 03:45 | |
| *** gongysh has quit IRC | 03:46 | |
| *** ducttape_ has joined #openstack-lbaas | 03:47 | |
| *** rcernin has joined #openstack-lbaas | 04:06 | |
| *** amotoki has quit IRC | 04:12 | |
| *** ducttape_ has quit IRC | 04:12 | |
| *** rcernin has quit IRC | 04:13 | |
| *** gcheresh_ has joined #openstack-lbaas | 04:23 | |
| *** amotoki has joined #openstack-lbaas | 04:30 | |
| *** amotoki has quit IRC | 04:48 | |
| *** chlong has quit IRC | 04:49 | |
| *** amotoki has joined #openstack-lbaas | 04:53 | |
| *** ducttape_ has joined #openstack-lbaas | 05:13 | |
| *** ducttape_ has quit IRC | 05:18 | |
| *** amotoki has quit IRC | 05:19 | |
| *** aojea has joined #openstack-lbaas | 05:21 | |
| *** gongysh has joined #openstack-lbaas | 05:27 | |
| *** aojea has quit IRC | 05:30 | |
| *** amotoki has joined #openstack-lbaas | 05:32 | |
| *** pcaruana has joined #openstack-lbaas | 06:08 | |
| *** krypto has joined #openstack-lbaas | 06:11 | |
| openstackgerrit | ZhaoBo proposed openstack/octavia master: Support QoS policy apply on vip port https://review.openstack.org/441912 | 06:14 |
| *** kobis has joined #openstack-lbaas | 06:17 | |
| *** rcernin has joined #openstack-lbaas | 06:31 | |
| *** krypto has quit IRC | 06:32 | |
| reedip | bzhao : review comments in the patch ^^ | 06:33 |
| *** krypto has joined #openstack-lbaas | 06:33 | |
| reedip | some are nits, some arent :) | 06:33 |
| *** sticker has quit IRC | 06:37 | |
| *** voelzmo has joined #openstack-lbaas | 06:40 | |
| *** ducttape_ has joined #openstack-lbaas | 06:44 | |
| *** sputnik13 has quit IRC | 06:44 | |
| *** voelzmo has quit IRC | 06:45 | |
| *** voelzmo has joined #openstack-lbaas | 06:45 | |
| *** sputnik13 has joined #openstack-lbaas | 06:45 | |
| *** ducttape_ has quit IRC | 06:48 | |
| *** tesseract has joined #openstack-lbaas | 06:52 | |
| *** voelzmo has quit IRC | 06:55 | |
| *** voelzmo has joined #openstack-lbaas | 06:56 | |
| *** voelzmo has quit IRC | 07:16 | |
| *** aojea has joined #openstack-lbaas | 07:24 | |
| *** voelzmo has joined #openstack-lbaas | 07:26 | |
| *** voelzmo has quit IRC | 07:35 | |
| *** aojea_ has joined #openstack-lbaas | 08:01 | |
| *** sanfern has joined #openstack-lbaas | 08:03 | |
| *** aojea has quit IRC | 08:04 | |
| *** ducttape_ has joined #openstack-lbaas | 08:14 | |
| *** ducttape_ has quit IRC | 08:19 | |
| *** voelzmo has joined #openstack-lbaas | 08:21 | |
| *** ianychoi has quit IRC | 08:24 | |
| *** ianychoi has joined #openstack-lbaas | 08:25 | |
| *** voelzmo has quit IRC | 08:26 | |
| *** reedip has quit IRC | 08:41 | |
| openstackgerrit | Nir Magnezi proposed openstack/octavia master: (WIP) Auto detect haproxy user_group https://review.openstack.org/429398 | 08:52 |
| *** krypto has quit IRC | 09:07 | |
| *** krypto has joined #openstack-lbaas | 09:08 | |
| bzhao | reedip_, thank you for review.:). We are moving our physical servers to the new place. Ah, so tired..I will update the patch soon. Thanks again. | 09:15 |
| *** bcafarel has quit IRC | 09:19 | |
| *** bcafarel has joined #openstack-lbaas | 09:27 | |
| *** reedip has joined #openstack-lbaas | 09:34 | |
| *** ducttape_ has joined #openstack-lbaas | 09:45 | |
| *** ducttape_ has quit IRC | 09:49 | |
| openstackgerrit | ZhaoBo proposed openstack/octavia master: Support QoS policy apply on vip port https://review.openstack.org/441912 | 09:52 |
| *** bbzhao has joined #openstack-lbaas | 09:52 | |
| *** voelzmo has joined #openstack-lbaas | 10:10 | |
| *** voelzmo has quit IRC | 10:14 | |
| *** pcaruana|afk| has joined #openstack-lbaas | 10:21 | |
| *** pcaruana|afk| has quit IRC | 10:23 | |
| *** pcaruana has quit IRC | 10:24 | |
| *** pcaruana has joined #openstack-lbaas | 10:25 | |
| *** amotoki has quit IRC | 10:26 | |
| *** gongysh has quit IRC | 10:35 | |
| *** voelzmo has joined #openstack-lbaas | 10:38 | |
| *** gongysh has joined #openstack-lbaas | 10:41 | |
| *** gongysh has quit IRC | 11:37 | |
| *** amotoki has joined #openstack-lbaas | 11:38 | |
| *** gongysh has joined #openstack-lbaas | 11:41 | |
| *** gongysh has quit IRC | 11:47 | |
| *** gongysh has joined #openstack-lbaas | 12:00 | |
| *** gongysh has quit IRC | 12:05 | |
| openstackgerrit | Guoqiang Ding proposed openstack/neutron-lbaas master: Choose correct log option by listener protocol https://review.openstack.org/453612 | 12:12 |
| *** ducttape_ has joined #openstack-lbaas | 12:26 | |
| *** amotoki has quit IRC | 12:41 | |
| *** krypto has quit IRC | 12:44 | |
| *** ducttape_ has quit IRC | 12:45 | |
| *** ducttape_ has joined #openstack-lbaas | 12:45 | |
| -openstackstatus- NOTICE: The Gerrit service on http://review.openstack.org is being restarted to address hung remote replication tasks, and should return to an operable state momentarily | 12:51 | |
| *** amotoki has joined #openstack-lbaas | 12:56 | |
| *** links has quit IRC | 13:08 | |
| *** catintheroof has joined #openstack-lbaas | 13:10 | |
| *** catintheroof has quit IRC | 13:10 | |
| *** catintheroof has joined #openstack-lbaas | 13:11 | |
| *** ducttape_ has quit IRC | 13:12 | |
| *** chlong has joined #openstack-lbaas | 13:17 | |
| *** ipsecguy_ is now known as ipsecguy | 13:31 | |
| *** ducttape_ has joined #openstack-lbaas | 13:36 | |
| *** krypto has joined #openstack-lbaas | 13:43 | |
| *** krypto has quit IRC | 13:48 | |
| *** krypto has joined #openstack-lbaas | 13:49 | |
| *** krypto has quit IRC | 13:49 | |
| *** krypto has joined #openstack-lbaas | 13:49 | |
| *** ducttape_ has quit IRC | 14:01 | |
| *** mdavidson has quit IRC | 14:18 | |
| *** ipsecguy_ has joined #openstack-lbaas | 14:29 | |
| *** ipsecguy has quit IRC | 14:31 | |
| *** krypto has quit IRC | 14:34 | |
| *** gcheresh_ has quit IRC | 14:35 | |
| *** krypto has joined #openstack-lbaas | 14:35 | |
| *** krypto has quit IRC | 14:35 | |
| *** krypto has joined #openstack-lbaas | 14:35 | |
| *** armax has joined #openstack-lbaas | 14:41 | |
| nmagnezi | rm_work, i'm getting close with the user_group removal on agent side.. | 14:54 |
| nmagnezi | :) | 14:54 |
| *** voelzmo has quit IRC | 15:01 | |
| *** chlong has quit IRC | 15:03 | |
| *** chlong has joined #openstack-lbaas | 15:04 | |
| *** fnaval has joined #openstack-lbaas | 15:04 | |
| *** fnaval_ has joined #openstack-lbaas | 15:06 | |
| *** armax has quit IRC | 15:07 | |
| *** fnaval__ has joined #openstack-lbaas | 15:08 | |
| *** ducttape_ has joined #openstack-lbaas | 15:08 | |
| *** fnaval has quit IRC | 15:10 | |
| *** fnaval has joined #openstack-lbaas | 15:10 | |
| *** fnaval___ has joined #openstack-lbaas | 15:12 | |
| *** fnaval_ has quit IRC | 15:12 | |
| *** fnaval_ has joined #openstack-lbaas | 15:13 | |
| *** fnaval__ has quit IRC | 15:14 | |
| *** fnaval__ has joined #openstack-lbaas | 15:15 | |
| *** fnaval has quit IRC | 15:16 | |
| *** fnaval has joined #openstack-lbaas | 15:17 | |
| *** fnaval___ has quit IRC | 15:17 | |
| *** fnaval_ has quit IRC | 15:18 | |
| *** fnaval_ has joined #openstack-lbaas | 15:18 | |
| *** fnaval___ has joined #openstack-lbaas | 15:20 | |
| *** fnaval__ has quit IRC | 15:20 | |
| *** chlong has quit IRC | 15:21 | |
| *** fnaval has quit IRC | 15:23 | |
| *** fnaval_ has quit IRC | 15:24 | |
| *** kobis has quit IRC | 15:26 | |
| *** ducttape_ has quit IRC | 15:34 | |
| *** chlong has joined #openstack-lbaas | 15:35 | |
| openstackgerrit | Sindhu Devale proposed openstack/octavia master: Fix import in v2 pool types https://review.openstack.org/453721 | 15:48 |
| *** aojea_ has quit IRC | 15:49 | |
| *** gcheresh_ has joined #openstack-lbaas | 15:52 | |
| *** ducttape_ has joined #openstack-lbaas | 15:59 | |
| *** ducttape_ has quit IRC | 16:15 | |
| *** cody-somerville has joined #openstack-lbaas | 16:16 | |
| *** armax has joined #openstack-lbaas | 16:17 | |
| *** armax has quit IRC | 16:27 | |
| *** gcheresh_ has quit IRC | 16:28 | |
| *** amotoki has quit IRC | 16:29 | |
| openstackgerrit | Merged openstack/octavia master: Align Octavia API to n-lbaasv2 for L7Policy https://review.openstack.org/406328 | 16:29 |
| *** armax has joined #openstack-lbaas | 16:31 | |
| *** cody-somerville has quit IRC | 16:31 | |
| *** tesseract has quit IRC | 16:43 | |
| *** armax has quit IRC | 16:43 | |
| openstackgerrit | Merged openstack/octavia master: Align Octavia API to n-lbaasv2 for L7Rules https://review.openstack.org/406336 | 16:45 |
| *** _ducttape_ has joined #openstack-lbaas | 16:59 | |
| *** sanfern has quit IRC | 17:18 | |
| *** _ducttape_ has quit IRC | 17:18 | |
| *** sanfern has joined #openstack-lbaas | 17:18 | |
| *** sanfern has quit IRC | 17:22 | |
| *** krypto has quit IRC | 17:28 | |
| johnsom | eandersson So I loaded up 200 load balancers into the DB and ran a 'neutron lbaas-loadbalancer-list' | 17:29 |
| johnsom | real 0m17.172s | 17:29 |
| johnsom | user 0m0.804s | 17:29 |
| johnsom | sys 0m0.108s | 17:29 |
| *** krypto has joined #openstack-lbaas | 17:29 | |
| johnsom | I suspect a good chunk of that is the client formatting. I'm going to do a native API call and see what I get. | 17:29 |
| johnsom | This is on a dev VM and not real hardware as well... | 17:30 |
| xgerman | johnsom remember when we did load tests at the company which shall not be named. I think we never had issues with list | 17:31 |
| eandersson | We are setting this up in the lab, with 64 LBs we are hitting 13s. | 17:31 |
| johnsom | xgerman yeah, I mentioned that yesterday. | 17:31 |
| eandersson | This is on A LOT slower hardware though. | 17:31 |
| johnsom | Granted, going through the layers of neutron probably don't help here... | 17:31 |
| xgerman | also the LB vms drain memory so if you try to pack that all on one box it will choke | 17:32 |
| xgerman | aka controller and compute node | 17:32 |
| johnsom | Yeah, these are "ghost LBs" Just the required database records for the API to do it's thing. | 17:34 |
| johnsom | Hmm, yeah, the API call is about the same as the client, so I guess formatting wasn't that bad. | 17:34 |
| johnsom | real 0m16.463s | 17:35 |
| xgerman | maybe if it has to traverse the network? | 17:36 |
| xgerman | aka goes from one controller node to the other (and not to speak of the load balancer people employ on the REST API) | 17:36 |
| xgerman | so you could easily get two or three roundtrips | 17:37 |
| johnsom | 400 LBs: real 0m32.477s | 17:37 |
| johnsom | So, kind of linear | 17:37 |
| johnsom | The actual DB query is less than a second. It's got to be either the sqlalchemy translation or something in the neutron layers.... | 17:38 |
| eandersson | Yea - looks like sqlachemy translation | 17:39 |
| *** armax has joined #openstack-lbaas | 17:39 | |
| eandersson | But going to dig deeper into that today. | 17:39 |
| xgerman | thanks — let us know | 17:40 |
| eandersson | We are running Mitaka btw | 17:41 |
| xgerman | uh, oh | 17:44 |
| eandersson | Couldn't find any major improvements on this code path though in master. | 17:44 |
| xgerman | yeah, we kept listing lbs pretty stable | 17:45 |
| johnsom | Yeah, in neutorn-lbaas that hasn't changed in a long time. I see the neutron process pegging out a CPU core, so.... | 17:45 |
| johnsom | Kind of curious how our new octavia API performs now. Since I have already trashed this devstack I might try it | 17:46 |
| *** asdf__ has quit IRC | 17:47 | |
| eandersson | Yea - I see a core pegged as well. | 17:47 |
| eandersson | > python-sqlalchemy-1.0.11-1.el7.x86_64 | 17:48 |
| eandersson | Going to try a newer version of sqlalchemy in the lab as well to rule that out. | 17:48 |
| johnsom | Well, the good news is our new Octavia API is a lot faster.... | 17:54 |
| johnsom | 400 LBs via Octavia v2 API: real 0m3.524s | 17:54 |
| *** gcheresh_ has joined #openstack-lbaas | 17:54 | |
| eandersson | Where is the Octacvia v2 Api? Is it different from Neutron-LBAAS v2 API? :D | 18:01 |
| johnsom | It is a compatible superset of the LBaaS v2 API. We are merging neutron-lbaas into octavia. | 18:02 |
| eandersson | Yep, read about that. Excited. | 18:02 |
| johnsom | http://docs-draft.openstack.org/57/438757/10/check/gate-octavia-api-ref/125623d//api-ref/build/html/ | 18:03 |
| johnsom | It's WIP | 18:03 |
| *** ducttape_ has joined #openstack-lbaas | 18:09 | |
| *** armax has quit IRC | 18:09 | |
| *** armax has joined #openstack-lbaas | 18:18 | |
| *** ducttape_ has quit IRC | 18:22 | |
| openstackgerrit | Shashank Kumar Shankar proposed openstack/octavia master: Run Octavia API in a WSGI server https://review.openstack.org/440934 | 18:28 |
| *** harlowja has quit IRC | 18:33 | |
| *** krypto has quit IRC | 18:35 | |
| *** krypto has joined #openstack-lbaas | 18:35 | |
| *** cody-somerville has joined #openstack-lbaas | 18:36 | |
| *** cody-somerville has quit IRC | 18:36 | |
| *** cody-somerville has joined #openstack-lbaas | 18:36 | |
| *** krypto has quit IRC | 18:52 | |
| *** cody-somerville has quit IRC | 18:56 | |
| *** cody-somerville has joined #openstack-lbaas | 18:57 | |
| *** csomerville has joined #openstack-lbaas | 19:03 | |
| *** cody-somerville has quit IRC | 19:03 | |
| *** armax has quit IRC | 19:08 | |
| *** csomerville has quit IRC | 19:08 | |
| *** csomerville has joined #openstack-lbaas | 19:09 | |
| *** ducttape_ has joined #openstack-lbaas | 19:12 | |
| *** csomerville has quit IRC | 19:14 | |
| *** cody-somerville has joined #openstack-lbaas | 19:16 | |
| *** cody-somerville has quit IRC | 19:16 | |
| *** cody-somerville has joined #openstack-lbaas | 19:16 | |
| *** cody-somerville has quit IRC | 19:21 | |
| *** cody-somerville has joined #openstack-lbaas | 19:24 | |
| *** voelzmo has joined #openstack-lbaas | 19:25 | |
| *** csomerville has joined #openstack-lbaas | 19:28 | |
| *** cody-somerville has quit IRC | 19:29 | |
| openstackgerrit | Shashank Kumar Shankar proposed openstack/octavia master: Run Octavia API in a WSGI server https://review.openstack.org/440934 | 19:30 |
| *** csomerville has quit IRC | 19:33 | |
| *** csomerville has joined #openstack-lbaas | 19:35 | |
| *** csomerville has quit IRC | 19:41 | |
| *** cody-somerville has joined #openstack-lbaas | 19:42 | |
| *** ducttape_ has quit IRC | 19:43 | |
| *** cody-somerville has quit IRC | 19:47 | |
| *** csomerville has joined #openstack-lbaas | 19:48 | |
| *** voelzmo has quit IRC | 19:53 | |
| johnsom | Octavia meeting starting soon on #openstack-meeting-alt | 19:56 |
| *** harlowja has joined #openstack-lbaas | 19:58 | |
| *** pcaruana has quit IRC | 20:01 | |
| *** aojea has joined #openstack-lbaas | 20:11 | |
| openstackgerrit | Shashank Kumar Shankar proposed openstack/octavia master: Run Octavia API in a WSGI server https://review.openstack.org/440934 | 20:16 |
| *** armax has joined #openstack-lbaas | 20:20 | |
| *** ducttape_ has joined #openstack-lbaas | 20:43 | |
| *** rcernin has quit IRC | 20:45 | |
| *** ducttape_ has quit IRC | 20:48 | |
| *** gcheresh_ has quit IRC | 20:50 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/neutron-lbaas master: Updated from global requirements https://review.openstack.org/453894 | 21:05 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/octavia master: Updated from global requirements https://review.openstack.org/453897 | 21:07 |
| *** bbzhao has quit IRC | 21:22 | |
| *** bbzhao has joined #openstack-lbaas | 21:23 | |
| *** aojea has quit IRC | 21:53 | |
| *** catintheroof has quit IRC | 21:54 | |
| *** chlong has quit IRC | 22:02 | |
| eandersson | johnsom, xgerman the version of sqlalchemy seems to have a farily large impact | 22:05 |
| *** harlowja has quit IRC | 22:05 | |
| xgerman | interesting | 22:05 |
| eandersson | in my lab it takes ~25-30s with 1.0.12 and 160 LBs | 22:05 |
| johnsom | Better results with something newer? | 22:05 |
| eandersson | with 1.1.7 it takes about 14-15s | 22:05 |
| eandersson | This might be why you guys are getting different numbers than me. | 22:06 |
| eandersson | 1.0.12 is the recommended for Mitaka though | 22:07 |
| johnsom | Yep. I can't point to a specific reason for that other than it is what was tested. You might be able to run your environment with a newer version if you test it out | 22:09 |
| eandersson | 14s for 160 LBs is still slow though, but better :p | 22:11 |
| eandersson | This is on significantly slower hardware though | 22:11 |
| *** ducttape_ has joined #openstack-lbaas | 22:14 | |
| eandersson | Actually the recommended version for Ocata (1.0.17) is as slow as the recommended version for Mitaka | 22:15 |
| johnsom | I think the Pike recommendation is 1.1.9 | 22:16 |
| johnsom | https://github.com/openstack/requirements/blob/master/upper-constraints.txt#L127 | 22:17 |
| eandersson | Nice - as 1.1.x gives a nice boost in performance here | 22:17 |
| *** gongysh has joined #openstack-lbaas | 22:19 | |
| *** ducttape_ has quit IRC | 22:21 | |
| *** fnaval___ has quit IRC | 22:51 | |
| openstackgerrit | Nir Magnezi proposed openstack/octavia master: Auto detect haproxy user_group https://review.openstack.org/429398 | 22:53 |
| nmagnezi | johnsom, https://review.openstack.org/#/c/429398/31/octavia/amphorae/backends/agent/api_server/listener.py@124 <<-- that was a tricky one (works!) | 22:54 |
| johnsom | Ok, will look | 22:54 |
| eandersson | Wow - when doing a single list lbs and this code is hit recursively 3338 times https://github.com/openstack/neutron-lbaas/blob/stable/mitaka/neutron_lbaas/services/loadbalancer/data_models.py#L74 | 23:00 |
| johnsom | nmagnezi Tricky indeed. So with the multiple config file it doesn't ignore the existing setting? | 23:03 |
| *** gongysh has quit IRC | 23:20 | |
| *** harlowja has joined #openstack-lbaas | 23:25 | |
| *** ducttape_ has joined #openstack-lbaas | 23:28 | |
| eandersson | https://github.com/openstack/neutron-lbaas/blob/stable/mitaka/neutron_lbaas/services/loadbalancer/data_models.py#L84 | 23:30 |
| eandersson | This is the line that takes time. | 23:30 |
| *** ducttape_ has quit IRC | 23:39 | |
| *** ducttape_ has joined #openstack-lbaas | 23:40 | |
| *** ducttape_ has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!