| melwitt | sean-k-mooney: patch to fix the GMR collection that I tried to do that did not work https://review.opendev.org/c/openstack/devstack/+/1012780 cc fungi clarkb | 01:33 |
|---|---|---|
| melwitt | the script was sending the signal to itself too which made it die 😆 | 01:34 |
| melwitt | the proposed patch ^ has helpfully failed CI for unrelated reasons so you can actually inspect the logs to see it correctly generating GMR in n-cpu and c-vol logs for example | 01:34 |
| sean-k-mooney | you can also confirug gmr to use a file based trigger, that waht we are doing downstream now we touch /var/lib/nova to trigger gmr | 02:03 |
| sean-k-mooney | im not sure why the curretn process/sheell would have recved the signal | 02:06 |
| sean-k-mooney | "because pkill -f | 02:07 |
| sean-k-mooney | matches the parent bash process's own cmdline, which contains | 02:07 |
| sean-k-mooney | all the service names from the script." | 02:07 |
| sean-k-mooney | ah that would do it | 02:07 |
| sean-k-mooney | we could have filter out bas i guess as well but ok ingoreing sign usr2 for that bash invocation should be fine | 02:08 |
| *** mhen_ is now known as mhen | 02:17 | |
| opendevreview | Merged openstack/tempest master: Register project admins from pre-provisioned account types https://review.opendev.org/c/openstack/tempest/+/1012417 | 04:28 |
| ralonsoh | Hello folks, can you spend some minutes on this patch chain https://review.opendev.org/c/openstack/devstack/+/1011786? | 06:57 |
| ralonsoh | thanks in advance | 06:57 |
| *** elodilles is now known as elodilles_pto | 08:10 | |
| opendevreview | Friedrich Zahn proposed openstack/tempest master: Fix endpoint filtering for missing region https://review.opendev.org/c/openstack/tempest/+/1012505 | 08:35 |
| opendevreview | Juan Larriba proposed openstack/devstack-plugin-jaeger master: Restart keystone after enabling oslo.middleware tracing https://review.opendev.org/c/openstack/devstack-plugin-jaeger/+/1012820 | 08:50 |
| opendevreview | Juan Larriba proposed openstack/devstack-plugin-jaeger master: Restart keystone after enabling oslo.middleware tracing https://review.opendev.org/c/openstack/devstack-plugin-jaeger/+/1012820 | 08:51 |
| opendevreview | Juan Larriba proposed openstack/devstack-plugin-jaeger master: Restart keystone after enabling oslo.middleware tracing https://review.opendev.org/c/openstack/devstack-plugin-jaeger/+/1012820 | 09:48 |
| *** pdeore_ is now known as pdeore | 14:03 | |
| clarkb | melwitt: what a fun bug to fix | 18:57 |
| melwitt | right? haha | 19:29 |
| melwitt | thanks sean-k-mooney and clarkb | 19:30 |
| sean-k-mooney | i consdier a few other was to fix it but in the end what you propsoed seemed fine | 19:31 |
| sean-k-mooney | filterign to only ptython or expcing bash process ectra all felt more complext then the trap | 19:32 |
| sean-k-mooney | thansk for documenting why this happend properly in the commti message | 19:32 |
| sean-k-mooney | it really was not obvious looking at the code why it was happening | 19:32 |
| sean-k-mooney | at least not until you read the commit | 19:33 |
| opendevreview | Merged openstack/devstack master: Fix GMR collection task killing itself and running too late https://review.opendev.org/c/openstack/devstack/+/1012780 | 20:29 |
| opendevreview | Merged openstack/devstack master: Set cafile option by base keystoneauth helper https://review.opendev.org/c/openstack/devstack/+/977253 | 21:48 |
| clarkb | ok cool the guru meditation change landed | 22:00 |
| clarkb | frickler: we dropped dstat from devstack by default because performance copilot packaging did not reliably install the package on older versions of ubuntu (and fedora too iirc) | 22:34 |
| clarkb | we don't run tests on those distro releases anymore so I pushed https://review.opendev.org/c/openstack/devstack/+/1011544 to reenable dstat by default. The first test runs seem to have been happy with it. I've rechecked it just now to get more data | 22:34 |
| clarkb | frickler: but I'm wondering if you thnik we can land that change (or something like it) since it is early in the release cycle and that way we can get this data out of jobs again? | 22:35 |
| clarkb | and worst case we disable it again if it becomes flaky | 22:35 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!