| opendevreview | Ghanshyam Maan proposed openstack/cinder master: Enable SRBAC by default https://review.opendev.org/c/openstack/cinder/+/986472 | 03:08 |
|---|---|---|
| opendevreview | Adam Harwell proposed openstack/cinder master: image_utils: replace direct tpool usage with concurrency helpers https://review.opendev.org/c/openstack/cinder/+/995001 | 05:36 |
| opendevreview | Jyotsna Lothe proposed openstack/cinder master: New cinder driver for HPE Alletra MP https://review.opendev.org/c/openstack/cinder/+/969373 | 08:55 |
| gmaan | rosmaita: if i remember correctly, you showed interest for enforce_scopre work in rbac, i have done the code changes for enabling it by default and it is ready for review, please check https://review.opendev.org/c/openstack/cinder/+/986472 | 13:51 |
| gmaan | this one https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/NYJOVFQ6FTXL7DKSOFIQJMOMX7R7CXLE/ | 13:51 |
| opendevreview | Ghanshyam Maan proposed openstack/cinder-tempest-plugin master: Testing RBAC new defaults enabling https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/995065 | 13:57 |
| opendevreview | Ghanshyam Maan proposed openstack/cinder master: DNM: testing rbac new defaults https://review.opendev.org/c/openstack/cinder/+/995083 | 15:08 |
| opendevreview | Jon Bernard proposed openstack/cinder master: [Everpure] Add a test for previous QoS patch https://review.opendev.org/c/openstack/cinder/+/995089 | 15:44 |
| jbernard | simondodsley: ^ i added a test as a followup, feel free to squash it into your original if you want/have time | 15:46 |
| opendevreview | Simon Dodsley proposed openstack/cinder master: [Everpure] Fix QoS for py-pure-client 1.86.0+ https://review.opendev.org/c/openstack/cinder/+/994947 | 16:01 |
| opendevreview | Simon Dodsley proposed openstack/cinder master: [Everpure] Fix QoS for py-pure-client 1.86.0+ https://review.opendev.org/c/openstack/cinder/+/994947 | 16:33 |
| opendevreview | Simon Dodsley proposed openstack/cinder master: [Everpure] Fix QoS for py-pure-client 1.86.0+ https://review.opendev.org/c/openstack/cinder/+/994947 | 16:41 |
| opendevreview | Adam Harwell proposed openstack/cinder master: Add ASGI application entry point for Cinder Volume API https://review.opendev.org/c/openstack/cinder/+/995131 | 19:06 |
| opendevreview | Adam Harwell proposed openstack/cinder master: Add ASGI application entry point for Cinder Volume API https://review.opendev.org/c/openstack/cinder/+/995131 | 19:12 |
| opendevreview | Jon Bernard proposed openstack/cinder master: Skip format_inspector tests when qemu-img is not installed https://review.opendev.org/c/openstack/cinder/+/995139 | 19:22 |
| opendevreview | Jon Bernard proposed openstack/cinder master: Use futurist-based thread pool for backup manager https://review.opendev.org/c/openstack/cinder/+/995140 | 19:29 |
| jbernard | rosmaita: ^ that's the smallest change i need to the backup subdir to get full unit test success, still leaves the backup migration to be done | 19:35 |
| rm_work[m] | ah is that the same topic as https://review.opendev.org/c/openstack/cinder/+/977851/18 | 19:36 |
| rm_work[m] | I was just looking to help with this and ended up making a patch on top of that one | 19:36 |
| rm_work[m] | https://review.opendev.org/c/openstack/cinder/+/995001/1 | 19:36 |
| rm_work[m] | I was able to get full devstack pass with that stack | 19:37 |
| rm_work[m] | (with eventlet disabled) | 19:37 |
| jbernard | yeah, me too, the winner will be chosen during review i guess so we have some choices | 19:38 |
| rm_work[m] | hmm ok | 19:39 |
| rm_work[m] | I do not care TBH which wins | 19:39 |
| rm_work[m] | I just want one to merge so non-eventlet works :D | 19:39 |
| rm_work[m] | did you also handle the stuff in my patch? | 19:39 |
| jbernard | which patch? | 19:40 |
| rm_work[m] | the second one I linked | 19:40 |
| rm_work[m] | I stacked it on the first one I linked | 19:40 |
| jbernard | lemme look, i have about 9 patches to clean up and submit here | 19:40 |
| rm_work[m] | image_utils | 19:40 |
| rm_work[m] | ah heh no worries | 19:40 |
| rm_work[m] | I am working on some stuff separately in oslo.messaging that I wanted a full devstack run to pass on both eventlet / non-eventlet and cinder was the last blocker | 19:41 |
| jbernard | ahh, no - we will need that one - i didn't install qemu-img in my container, so i did the right thing and skipped all of those :) https://review.opendev.org/c/openstack/cinder/+/995139/1 | 19:42 |
| rm_work[m] | hah | 19:42 |
| rm_work[m] | I could also rebase it onto your CR | 19:43 |
| jbernard | i have a full patchset locally that fixes everything (minus the image-utils) | 19:43 |
| rm_work[m] | or you could take it over | 19:43 |
| rm_work[m] | I just decided to do the quick CR rather than coming here to complain :P | 19:43 |
| jbernard | no worries :) | 19:43 |
| rm_work[m] | but if you have the full context and are shepherding things through, I don't need to own that | 19:43 |
| jbernard | i need to clean things up a bit and get everything into gerrit | 19:43 |
| rm_work[m] | that CR was like 99% claude anyway | 19:44 |
| rm_work[m] | but it does pass devstack now :D | 19:44 |
| jbernard | yeah, i think ive got it | 19:44 |
| jbernard | well - unit tests passing anyway | 19:44 |
| jbernard | interesting, there are a few missing pieces for that to be possible in my env, i need to look at what devstack has done | 19:45 |
| jbernard | i also found a bug in oslo.service in the process | 19:45 |
| jbernard | and fixed some long-running tests in hpe and storwize while i was in there | 19:45 |
| jbernard | rm_work[m]: when you say passes, what command are you running? 'tox -e py3-threading'? | 19:47 |
| rm_work[m] | full dsvm | 19:48 |
| rm_work[m] | I can show you | 19:48 |
| rm_work[m] | https://review.opendev.org/c/openstack/oslo.messaging/+/994601 | 19:48 |
| rm_work[m] | https://zuul.opendev.org/t/openstack/build/5fde6d3292b64354ae4a3bde5a23be88 | 19:48 |
| rm_work[m] | this run was with eventlet turned off | 19:48 |
| rm_work[m] | and I know it really was off because in that codepath it would literally just lock-up the processes and fail if it was on | 19:49 |
| jbernard | ok, ill take a close look at that, thanks | 19:49 |
| rm_work[m] | it's kinda hacked together, I will need to make a real CR to devstack repo | 19:50 |
| rm_work[m] | I may do that now actually | 19:51 |
| jbernard | rm_work[m]: i see results for oslo-messaging, but im curious about how cinder's tests were run - i dont know where to look to find those | 19:53 |
| jbernard | rm_work[m]: we still have eventlet in our requirements, so it will get pulled in and imported | 19:53 |
| rm_work[m] | just standard devstack but with OS_CINDER_DISABLE_EVENTLET_PATCHING=true | 19:54 |
| jbernard | ok, that would be equivalent to py3-threading then | 19:54 |
| rm_work[m] | like I said, I am pretty sure it worked, because the tests thoroughly failed when eventlet was on, because the code it tests does bad thread stuff otherwise | 19:54 |
| rm_work[m] | without Depends-On to my patch: fail, with Depends-On: pass | 19:54 |
| rm_work[m] | though is there a c-bak or something | 19:55 |
| rm_work[m] | or does backup fall into something else | 19:55 |
| rm_work[m] | 'SYSTEMD_ENV_VARS["c-api"]': OS_CINDER_DISABLE_EVENTLET_PATCHING=true | 19:55 |
| rm_work[m] | 'SYSTEMD_ENV_VARS["c-sch"]': OS_CINDER_DISABLE_EVENTLET_PATCHING=true | 19:55 |
| rm_work[m] | 'SYSTEMD_ENV_VARS["c-vol"]': OS_CINDER_DISABLE_EVENTLET_PATCHING=true | 19:55 |
| rm_work[m] | that is all I have set | 19:55 |
| jbernard | there is if you enable it, and that would need to be added | 19:56 |
| jbernard | each service has to be told to not monkey patch (at the moment) | 19:56 |
| rm_work[m] | well it was obviously enabled by default, because it was being tested and the tests all failed because my threading caused the process to lock | 19:57 |
| jbernard | hmm, i clearly need to familiarize myself better with the dvsm environment | 19:57 |
| rm_work[m] | well, I am double-checking because maybe wires got crossed, it might have disabled those tests at some point while I was fixing it and not re-enabled, I am checking | 19:58 |
| jbernard | im confident in this: | 19:58 |
| jbernard | if you clone the cinder repo and add the pending gerrit patches related to eventlet-removal (topic), the py3-threading tox run comes up with about 10 failing tests | 19:59 |
| jbernard | which all appear to be valid failures | 19:59 |
| jbernard | i have patches that resolve those, but there does remain eventlet use that will pass the native threading env anyway, it seems | 20:00 |
| jbernard | mostly in cinder-backup | 20:00 |
| rm_work[m] | hmmm yeah per my investigation, the patch I linked / based my fixes on did take care of those functionally | 20:01 |
| rm_work[m] | tho it did still have some unit test failures | 20:01 |
| rm_work[m] | if you have better context on that I should probably just leave it to you TBH | 20:01 |
| jbernard | i have been focusing on the unit tests, haven't looked at functional yet | 20:01 |
| rm_work[m] | I'm just a passer-by ;) | 20:01 |
| jbernard | no worries, ill let you know once i get everything up and we can all give it a spin | 20:03 |
| rm_work[m] | yeah ping me whenever | 20:04 |
| opendevreview | Adam Harwell proposed openstack/cinder master: Add ASGI application entry point for Cinder Volume API https://review.opendev.org/c/openstack/cinder/+/995131 | 20:20 |
| opendevreview | Merged openstack/cinder master: [Everpure] Fix QoS for py-pure-client 1.86.0+ https://review.opendev.org/c/openstack/cinder/+/994947 | 21:24 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!