| opendevreview | sean mooney proposed openstack/devstack master: ignore dbcounter sub dirs https://review.opendev.org/c/openstack/devstack/+/898844 | 01:43 |
|---|---|---|
| opendevreview | sean mooney proposed openstack/devstack master: [DNM] testing centos9/rocky9 jobs with global venv enabled https://review.opendev.org/c/openstack/devstack/+/898845 | 01:43 |
| ykarel | clarkb, for that privsep-helper missing i believe it's not the same bug as ubuntu-jammy one, but something to do with the path where oslo.privsep is installed and it's path inclusion in exec_dirs for /etc/neutron/rootwrap.conf | 04:09 |
| ykarel | harsh seems to be out now | 04:09 |
| ykarel | and as with package installation it works it would have been in the path already included in exec_dirs | 04:10 |
| ykarel | but yes full devstack install logs would have some hint on what went wrong or some reproducible steps as ovn jobs in yoga 20.04 are working fine | 04:12 |
| sean-k-mooney | gmann: frickler so i made some progress on GLOBAL_VENV=true on centos | 11:57 |
| sean-k-mooney | basically the "bug" is related to pep 517, pbr and our custome wsgi_scripts fields | 11:58 |
| sean-k-mooney | if we pin to an older pip on centos where we dont try to build the wheel | 11:58 |
| sean-k-mooney | pbr will generate teh wsgi script correctly | 11:59 |
| sean-k-mooney | when we build the editable wheel we dont | 11:59 |
| sean-k-mooney | using the old pip breaks debian because python-ovs on that plathform needs it as they now require pep517 support to install | 12:00 |
| sean-k-mooney | i think if we remove the if here https://github.com/openstack/pbr/blob/master/pbr/packaging.py#L493-L498 | 12:01 |
| sean-k-mooney | and allways call self._make_wsgi_scripts_only(dist, executable) | 12:01 |
| sean-k-mooney | that might help fix this in pep517 mode? | 12:02 |
| sean-k-mooney | clarkb: ^ | 12:02 |
| clarkb | sean-k-mooney: the issue is bdist wheel is different than the editable wheel instruction? Maybe we should check for that instead? | 14:04 |
| clarkb | I'm not sure what the effect is for building those scripts for all build cases. Maybe we leak files into the build dir which isn't a big deal? | 14:05 |
| sean-k-mooney | clarkb: it also does not work in non editable mode fyi | 14:48 |
| sean-k-mooney | clarkb: the issue i think is but ya perhasp we need to check for a list of options | 14:49 |
| sean-k-mooney | i have not had time to do it but i was going to add some logging to see what its invoked with | 14:49 |
| clarkb | I'msurprised that building a non editable wheel doesn't go through that bdist wheel case | 14:50 |
| clarkb | maybe they aer using a new flag for pep517 which would explain it I guess | 14:50 |
| sean-k-mooney | ya i think its something like that | 14:50 |
| sean-k-mooney | a new flag | 14:50 |
| opendevreview | Ghanshyam proposed openstack/tempest master: Switch Neutron RBAC flag in testing https://review.opendev.org/c/openstack/tempest/+/899310 | 14:52 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!