| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2020-06-08 | |||
| 14:47:08 | sean-k-mooney | yep it is https://opendev.org/openstack/governance/src/branch/master/reference/runtimes/victoria.rst#user-content-python-runtimes-for-victoria | |
| 14:47:45 | sean-k-mooney | so 3.6 and 3.8 will be the required supported version once we swap form 18.04 | |
| 14:48:00 | sean-k-mooney | to 20.04 | |
| 14:48:57 | sean-k-mooney | frickler: the eventlet thing seams to be intermitent | |
| 14:49:08 | sean-k-mooney | i see the command are sometimes passing fine | |
| 14:52:05 | sean-k-mooney | frickler: given i deployed master i would apper that https://review.opendev.org/#/c/707474/ does not actully work | |
| 14:53:33 | frickler | sean-k-mooney: you might want to test with a revert, but I was assuming that it would be only incomplete and we would have much more issues without it | |
| 14:54:05 | sean-k-mooney | yes it likely incomplete but i can try that quickly. | |
| 14:54:18 | sean-k-mooney | me might need to do this in oslo as well | |
| 15:00:59 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: DNM: Test the state of VMware NSX 3pp CI https://review.opendev.org/734114 | |
| 19:19:45 | dansmith | sean-k-mooney: around? | |
| 19:28:08 | sean-k-mooney | dansmith: yes | |
| 19:28:18 | sean-k-mooney | just traging a networking issue how can i help | |
| 19:28:44 | dansmith | sean-k-mooney: I want to create a variant of the ceph job that has multiple glance stores, so some glance conf tweaking | |
| 19:28:56 | dansmith | sean-k-mooney: I think I need a devstack_local_conf section in the job def in .zuul.conf yeah? | |
| 19:29:18 | dansmith | I'm not quite sure what it needs to look like, can I pastebin something in a sec for you to look at? | |
| 19:29:38 | sean-k-mooney | so you likely need to set different values on the contoler/all-in-one node vs the compute only node | |
| 19:29:49 | sean-k-mooney | so you will need to set different value in different groups | |
| 19:30:07 | sean-k-mooney | dansmith: sure give me the linke and ill take a look | |
| 19:30:41 | dansmith | this can be single-node fwiw | |
| 19:30:58 | sean-k-mooney | with multiple stores? | |
| 19:31:01 | dansmith | yup | |
| 19:31:20 | sean-k-mooney | oh are you configuring mutile glance backends or something | |
| 19:32:02 | sean-k-mooney | i assume it cant advertises the single file store backend as multiple stores | |
| 19:32:20 | sean-k-mooney | or in this case ceph backends | |
| 19:32:52 | dansmith | one file and one ceph, but yeah you can have multiple file backends too AFAIK | |
| 19:33:25 | sean-k-mooney | ya im pretty sure you can have multipel file backedns | |
| 19:33:40 | sean-k-mooney | https://github.com/openstack/nova/blob/master/.zuul.yaml#L154-L161 | |
| 19:33:48 | sean-k-mooney | basically you are doing that | |
| 19:34:10 | sean-k-mooney | but replacen $NOVA_CPU_CONF with $GLANCE_CONF | |
| 19:34:32 | sean-k-mooney | then each "[section]" becomes "section:" | |
| 19:34:33 | dansmith | sean-k-mooney: https://pastebin.com/h8Z2hTQs | |
| 19:34:53 | dansmith | okay I thought so, but I saw some in that file that didn't look like section names I was expecting | |
| 19:35:02 | dansmith | so I thought maybe there was more to it | |
| 19:35:24 | sean-k-mooney | so the $WHATEVER | |
| 19:35:30 | sean-k-mooney | is being replaced by devstack | |
| 19:35:49 | dansmith | I figured | |
| 19:35:49 | sean-k-mooney | althoguh you are missing a bit | |
| 19:35:59 | sean-k-mooney | https://github.com/openstack/nova/blob/master/.zuul.yaml#L155 | |
| 19:36:08 | sean-k-mooney | you will notice its under post-config | |
| 19:36:24 | dansmith | ah okay thanks | |
| 19:36:39 | sean-k-mooney | so line 155 and 156 become [[post-config|$NOVA_CPU_CONF]] | |
| 19:37:48 | dansmith | ack | |
| 19:38:12 | sean-k-mooney | im sure you know this but the post-config section in devstack run after all the infrastucture serivces are start and after all the default config are generated but before any of the openstack services are started | |
| 19:38:28 | dansmith | yeah that's what I want | |
| 19:38:49 | sean-k-mooney | yep its both the latest point you can set it and in this casse the earlist that makes sense | |
| 19:39:01 | dansmith | sean-k-mooney: so what about the bit of parenting this from the existing job and replacing the existing thing under check to reference this... that look okay too? | |
| 19:39:50 | sean-k-mooney | yes although when im testing i normally disable the other check jobs untill i get it working | |
| 19:40:05 | dansmith | by just deleting them? | |
| 19:40:35 | sean-k-mooney | https://review.opendev.org/#/c/681771/2/.zuul.yaml | |
| 19:40:48 | sean-k-mooney | i normally coment them out but ya deleting works too | |
| 19:40:49 | dansmith | ack | |
| 19:40:52 | dansmith | smart :) | |
| 19:41:19 | sean-k-mooney | actully thats slight a bad example since i enabled noop and use the experimental queue to test | |
| 19:41:26 | sean-k-mooney | the expermiental queue is faster | |
| 19:41:39 | sean-k-mooney | so it will report back sooner | |
| 19:41:51 | sean-k-mooney | normaly i just use check with the other disabeld | |
| 19:42:11 | sean-k-mooney | the goal is the same jsut run the one job i care about and not waste ci resouces | |
| 19:42:33 | dansmith | ack, I will leave it in check because it's easier | |
| 19:42:58 | dansmith | gah, conflict in .zuul | |
| 19:43:04 | sean-k-mooney | but ya once you do that all you need to do is reference it by its new name and you should be good to go | |
| 19:43:22 | sean-k-mooney | ya that is the downside to commenting | |
| 19:44:08 | sean-k-mooney | by the way i normally comment out all template bar - check-requirements | |
| 19:44:46 | sean-k-mooney | but again that is not needed i just normally do that while im getting the job working to disable pep8 and the tox jobs | |
| 19:44:53 | openstackgerrit | Dan Smith proposed openstack/nova master: Plumb image import functionality through our glance module https://review.opendev.org/731550 | |
| 19:44:54 | openstackgerrit | Dan Smith proposed openstack/nova master: Make libvirt able to trigger a backend image copy when needed https://review.opendev.org/656998 | |
| 19:44:55 | openstackgerrit | Dan Smith proposed openstack/nova master: WIP: Add libvirt RBD configuration doc https://review.opendev.org/731577 | |
| 19:44:55 | openstackgerrit | Dan Smith proposed openstack/nova master: DNM: Try to make a glance multistore job https://review.opendev.org/734184 | |
| 19:45:06 | dansmith | sean-k-mooney: ^ | |
| 19:45:34 | dansmith | sean-k-mooney: thanks for your help. it's so much easier to just hint at you doing it, but I have to do it myself *sometimes* :) | |
| 19:45:56 | sean-k-mooney | you proably dont need the required-projects: nova but it wont break anything | |
| 19:46:05 | sean-k-mooney | you will get that form devstack via inheritance | |
| 19:46:12 | dansmith | I copied that from another job | |
| 19:46:19 | sean-k-mooney | ya its fine | |
| 19:46:25 | dansmith | blindly without knowing why.. seems silly in the nova tree | |
| 19:46:29 | sean-k-mooney | its already required by tempest or devstack | |
| 19:46:33 | dansmith | ack | |
| 19:47:43 | sean-k-mooney | so what that does is it adds nova to LIBS_FROM_GIT | |
| 19:47:58 | sean-k-mooney | which for libs has the sideffect of make depens-on work | |
| 19:48:19 | sean-k-mooney | but its also used for zuul to know which poject to allow depens on to work with | |
| 19:48:24 | sean-k-mooney | in general that is | |
| 19:48:53 | sean-k-mooney | so if you want depens on to work say with ooo then somewhere in the job inheritance you would need a reqiured projects | |
| 19:49:07 | sean-k-mooney | so that he zuul cloner can prepare it properly | |
| 19:49:43 | dansmith | ah | |
| 19:49:49 | dansmith | sounds like one of those things it's easy to forget until after a few hours of tearing out hair wondering why | |
| 19:49:55 | sean-k-mooney | you normally dont need to know that | |
| 19:50:04 | sean-k-mooney | yep exactly | |
| 19:50:50 | sean-k-mooney | for os-vif for example if you dont add it it will use the latest version allowed by upper-constraints.txt instead of master | |
| 19:51:05 | sean-k-mooney | which is normaly the correct behavior unless you are creating an os-vif job | |
| 19:52:09 | dansmith | we should just squash all the projects back into one tree obviously.. this would be so much easier ;) | |
| 19:52:42 | sean-k-mooney | hehe it works for the kernel but i like gerrit not exploding | |
| 19:52:57 | sean-k-mooney | although to be fair im not sure if would have an issue with that | |
| 19:53:12 | dansmith | well, I'm joking of course | |
| 19:53:14 | sean-k-mooney | i know that gitea would cry whoever | |
| 19:53:17 | dansmith | but sometimes ... it'd be easier | |
| 19:53:27 | sean-k-mooney | sometimes yes | |
| 19:53:54 | sean-k-mooney | i know that neutron-lib can be a pain for the neutron folks in that regard | |
| 19:54:04 | dansmith | aight, well, time to go do something else while that runs so I can see how close it is | |
| 19:54:06 | dansmith | thanks again | |
| 19:54:09 | sean-k-mooney | its partly why tempest lib is now dead | |