| *** mhen_ is now known as mhen | 01:54 | |
| opendevreview | Hervé Beraud proposed openstack/oslo.db master: Add MariaDB error 1020 handling as DBConsistencyError subclass of DBDeadlock https://review.opendev.org/c/openstack/oslo.db/+/955189 | 08:24 |
|---|---|---|
| opendevreview | Dmitry Tantsur proposed openstack/futurist master: Trivial: avoid noqa in __init__.py https://review.opendev.org/c/openstack/futurist/+/955374 | 13:21 |
| opendevreview | Dmitry Tantsur proposed openstack/futurist master: WIP: Add DynamicThreadPoolExecutor that resizes itself https://review.opendev.org/c/openstack/futurist/+/955217 | 15:33 |
| dtantsur | TheJulia: could you retest ^^^ at your convenience? This is the new approach relying on busy threads | 15:33 |
| opendevreview | Artem Goncharov proposed openstack/pycadf master: Add ruff https://review.opendev.org/c/openstack/pycadf/+/952366 | 15:34 |
| TheJulia | dtantsur: sure! I’ll try to kick it off in about an hour | 15:36 |
| TheJulia | dtantsur: so far so good, I'll let you know if the spread is any different on the runs | 16:04 |
| opendevreview | Merged openstack/pycadf master: Drop explicit dependency on python-subunit https://review.opendev.org/c/openstack/pycadf/+/952246 | 16:13 |
| TheJulia | dtantsur: yeah... I think that might have been a huge regression from the prior. 16:11 for the first power sync pass, 16:38 for the second. Also hanging at 26 threads locally. I'll let it keep looping to see if that time comes down | 16:48 |
| TheJulia | 17:05 was the third pass :\ | 17:26 |
| dtantsur | TheJulia: you may need to set parameters differently though | 18:03 |
| * TheJulia takes a deeper look | 18:04 | |
| dtantsur | Granted, this one may create new threads earlier and drop them later | 18:04 |
| dtantsur | There is debug logging now that you can use | 18:04 |
| * dtantsur gets back into the evening business | 18:05 | |
| dtantsur | TheJulia: I have a combined idea that we can try on Monday: instead of checking n_busy_threads / n_threads, we can also use a part of the previous logic and check (n_busy_threads + queue_size) / n_threads. That will account for work that is about to get scheduled. | 18:19 |
| dtantsur | Then we can play with thresholds to see how quickly the threads get created/reaped | 18:19 |
| TheJulia | yeah, I started it with a minimum worker pool of 10, and default thresholds, I'm really wondering why it is so much slower, but I did revert the rejection setting back to 1 as you indicated you were fixing the logic | 18:20 |
| dtantsur | meanwhile, oslo folks, could you please remove a pep8-incompatible recommendation? https://review.opendev.org/c/openstack/futurist/+/955217/2/futurist/_futures.py | 18:20 |
| *** haleyb is now known as haleyb|out | 18:33 | |
| hberaud[m] | dtantsur: we have a look early next week | 19:02 |
| hberaud[m] | s/we/will/ | 19:02 |
| -opendevstatus- NOTICE: The Gerrit service on review.opendev.org will be offline briefly for a configuration and version update, but should return to service momentarily | 20:07 | |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!