| *** slaweq has quit IRC | 00:12 | |
| *** slaweq has joined #openstack-rally | 00:13 | |
| *** irclogbot_2 has quit IRC | 00:39 | |
| *** wangy_ has joined #openstack-rally | 01:18 | |
| *** aojea has joined #openstack-rally | 01:23 | |
| *** aojea has quit IRC | 01:27 | |
| *** threestrands has joined #openstack-rally | 02:30 | |
| *** wangy_ has quit IRC | 03:01 | |
| *** noama has joined #openstack-rally | 04:54 | |
| *** agopi has joined #openstack-rally | 05:08 | |
| *** wznoinsk has quit IRC | 05:46 | |
| *** linuxaddicts has joined #openstack-rally | 06:08 | |
| *** threestrands has quit IRC | 06:23 | |
| *** linuxaddicts has quit IRC | 06:23 | |
| *** aojea has joined #openstack-rally | 06:27 | |
| *** aojea has quit IRC | 06:31 | |
| *** pcaruana has joined #openstack-rally | 07:36 | |
| *** e0ne has joined #openstack-rally | 07:49 | |
| *** kbaikov has quit IRC | 09:13 | |
| slaweq | andreykurilin: hi | 09:24 |
|---|---|---|
| slaweq | andreykurilin: I was checking this rally and osprofiler integration in neutron job and unfortunatelly it looks that it's too slow to generate those reports - even if I used 3h timeout for a job it got timeout :/ | 09:25 |
| *** linuxaddicts has joined #openstack-rally | 09:31 | |
| andreykurilin | slaweq: yes, I saw it. It happens since each iteration of each workload has a separate trace. Now I'm thinking how to fix it and optimize | 09:37 |
| slaweq | andreykurilin: that would be great, I can try to help with that but I don't know rally at all so may be hard for me :) | 09:38 |
| andreykurilin | slaweq: do you want to check a specific workload now or you would like to do it permanently? I mean that I can propose a temporary hack if you want to check something quickly and that think about proper solution | 09:39 |
| slaweq | andreykurilin: we want to add it permanently to be able to track some API slowdowns in neutron | 09:47 |
| *** e0ne has quit IRC | 10:02 | |
| *** e0ne has joined #openstack-rally | 10:04 | |
| *** linuxaddicts has quit IRC | 10:27 | |
| *** e0ne has quit IRC | 10:45 | |
| *** e0ne has joined #openstack-rally | 10:54 | |
| *** aojea_ has joined #openstack-rally | 11:18 | |
| *** aojea_ has quit IRC | 11:23 | |
| *** kbaikov has joined #openstack-rally | 11:27 | |
| *** e0ne has quit IRC | 12:04 | |
| *** e0ne has joined #openstack-rally | 12:08 | |
| *** e0ne has quit IRC | 12:09 | |
| *** e0ne has joined #openstack-rally | 12:27 | |
| andreykurilin | slaweq: I do not have a good solution right now. need a bit time to think | 12:51 |
| slaweq | andreykurilin: sure, thx for taking care of it | 12:53 |
| slaweq | andreykurilin: one more question, maybe You will know - do You think that it might be faster if we would use different backend in osprofiler? | 12:53 |
| andreykurilin | slaweq: I do not thing that the problem in the backend. Let me try to explain what is happening - your rally task contains several workloads. In total they have >500 iterations (one execution of the scenario). Each iteration has own trace-id which is mapped to the whole trace. rally html report fetches all these traces and tries to put them into rally report. You mentioned that reddis db had size ~2gb and I can guess that all these 2gb is these | 13:00 |
| andreykurilin | traces which means that rally tries to fetch 2gb of data in real-time (possible we have ram limit here) and put them into html report which in result will be huge and unreadable | 13:00 |
| andreykurilin | slaweq: ,aybe we can add more debug messages to prove this theory | 13:01 |
| andreykurilin | If I'm not mistaken, we need to rethink the integration of osprofiler reports into rally's. store them separately and lazy-load... | 13:04 |
| slaweq | andreykurilin: yes, I agree that this what happens there. I just thought that maybe using some other backend could speed up generation of html of each request in osprofiler, but probably even if some other backend would be faster it wouldn't be significant improvement :/ | 13:07 |
| andreykurilin | slaweq: in general, the single traceback should not be huge. It is just simple dict with ~100 timeseries items... so even 500 iterations should not be so huge | 13:08 |
| andreykurilin | more debug messages should help to understand the issue | 13:08 |
| slaweq | andreykurilin: ok | 13:10 |
| openstackgerrit | Andrey Kurilin proposed openstack/rally-openstack master: DNM: add some debug-mesages https://review.openstack.org/616191 | 13:14 |
| openstackgerrit | Andrey Kurilin proposed openstack/rally-openstack master: DNM: add some debug-mesages https://review.openstack.org/616191 | 13:16 |
| openstackgerrit | Andrey Kurilin proposed openstack/rally-openstack master: DNM: add some debug-mesages https://review.openstack.org/616191 | 13:17 |
| andreykurilin | slaweq: Can you change `depends-on ` of your patch to https://review.openstack.org/616191 patch. Hope it will give more data | 13:18 |
| andreykurilin | ? | 13:18 |
| *** e0ne has quit IRC | 13:41 | |
| *** pcaruana has quit IRC | 13:45 | |
| *** e0ne has joined #openstack-rally | 13:45 | |
| *** agopi has quit IRC | 13:47 | |
| *** pcaruana has joined #openstack-rally | 14:00 | |
| *** aojea_ has joined #openstack-rally | 14:04 | |
| *** e0ne has quit IRC | 14:13 | |
| *** e0ne has joined #openstack-rally | 14:14 | |
| *** agopi has joined #openstack-rally | 14:17 | |
| *** rh-jelabarre has joined #openstack-rally | 14:21 | |
| openstackgerrit | Chuck Short proposed openstack/rally-openstack master: Add scenarios for octavia https://review.openstack.org/615922 | 14:24 |
| *** pcaruana has quit IRC | 14:33 | |
| *** pcaruana has joined #openstack-rally | 14:34 | |
| openstackgerrit | Merged openstack/rally-openstack master: Fix fetching osprofiler trace-info https://review.openstack.org/615889 | 14:53 |
| *** aojea_ has quit IRC | 15:04 | |
| *** aojea_ has joined #openstack-rally | 15:30 | |
| slaweq | andreykurilin: done | 15:34 |
| *** aojea_ has quit IRC | 15:35 | |
| *** aojea_ has joined #openstack-rally | 15:35 | |
| *** aojea_ has quit IRC | 15:35 | |
| *** aojea_ has joined #openstack-rally | 15:36 | |
| *** aojea_ has quit IRC | 15:56 | |
| *** aojea_ has joined #openstack-rally | 15:57 | |
| *** aojea_ has quit IRC | 16:06 | |
| *** agopi is now known as agopi|lunch | 16:14 | |
| openstackgerrit | Chuck Short proposed openstack/rally-openstack master: Add scenarios for octavia https://review.openstack.org/615922 | 16:19 |
| *** pcaruana has quit IRC | 16:38 | |
| *** e0ne has quit IRC | 16:44 | |
| *** e0ne has joined #openstack-rally | 16:44 | |
| *** e0ne has quit IRC | 16:44 | |
| *** rh-jelabarre has quit IRC | 17:35 | |
| *** noama has quit IRC | 17:46 | |
| *** agopi|lunch is now known as agopi | 18:15 | |
| *** aojea_ has joined #openstack-rally | 18:26 | |
| *** e0ne has joined #openstack-rally | 18:36 | |
| *** e0ne has quit IRC | 18:38 | |
| *** irclogbot_2 has joined #openstack-rally | 18:46 | |
| *** rh-jelabarre has joined #openstack-rally | 18:52 | |
| *** rh-jelabarre has quit IRC | 18:52 | |
| *** rh-jelabarre has joined #openstack-rally | 18:52 | |
| *** aojea_ has quit IRC | 18:57 | |
| *** rh-jelabarre has quit IRC | 19:18 | |
| *** e0ne has joined #openstack-rally | 19:24 | |
| *** rh-jelabarre has joined #openstack-rally | 19:29 | |
| *** e0ne has quit IRC | 19:31 | |
| *** aojea_ has joined #openstack-rally | 19:49 | |
| *** aojea_ has quit IRC | 20:21 | |
| *** aojea_ has joined #openstack-rally | 20:21 | |
| *** aojea_ has quit IRC | 20:43 | |
| *** aojea_ has joined #openstack-rally | 20:46 | |
| *** aojea_ has quit IRC | 20:56 | |
| *** aojea_ has joined #openstack-rally | 20:58 | |
| *** aojea_ has quit IRC | 21:05 | |
| *** aojea_ has joined #openstack-rally | 21:06 | |
| *** rh-jelabarre has quit IRC | 21:10 | |
| *** rh-jelabarre has joined #openstack-rally | 21:28 | |
| *** rh-jelabarre has quit IRC | 22:54 | |
| *** slaweq has quit IRC | 23:01 | |
| *** aojea_ has quit IRC | 23:24 | |
| *** aojea_ has joined #openstack-rally | 23:27 | |
| *** aojea_ has quit IRC | 23:32 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!