digitalsimboja | @maysams, @ltomasbo: please assist to resolve | 06:53 |
---|---|---|
digitalsimboja | https://paste.opendev.org/show/807743/ | 06:54 |
ltomasbo | digitalsimboja, what did you execute there? | 06:54 |
digitalsimboja | Good morning | 06:54 |
digitalsimboja | this | 06:54 |
digitalsimboja | stack@outreachy:~/tempest$ tempest run -r 'kuryr-tempest-plugin.*' | 06:54 |
ltomasbo | looks like you run out of disk space | 06:56 |
gryf | ltomasbo, I also suspect that. | 06:56 |
digitalsimboja | oops! | 06:57 |
gryf | digitalsimboja, df -h | 06:57 |
digitalsimboja | okay | 06:57 |
digitalsimboja | https://paste.opendev.org/show/807744/ | 06:58 |
ltomasbo | yep: /dev/vda1 39G 39G 0 100% / | 06:58 |
ltomasbo | digitalsimboja, probably just removing some of the log files would help there | 06:59 |
digitalsimboja | moment.. | 06:59 |
ltomasbo | yep, /var/log is taking up to 28G | 06:59 |
ltomasbo | https://paste.opendev.org/show/807746/ | 07:00 |
ltomasbo | you can probably just remove the syslog.1 file and get things recovered | 07:01 |
digitalsimboja | checking.. | 07:01 |
digitalsimboja | okay I removed the syslog files | 07:03 |
digitalsimboja | do I need to remove /dev/vdal files too? | 07:03 |
gryf | no! | 07:06 |
ltomasbo | and probably you should have not removed the syslog but set its size to 0, lets see if it gets regenerated | 07:07 |
digitalsimboja | now I am getting this^ | 07:09 |
digitalsimboja | https://paste.opendev.org/show/807747/ | 07:09 |
digitalsimboja | I ran stestr init | 07:10 |
digitalsimboja | yet | 07:10 |
ltomasbo | digitalsimboja, this is the same as yesterday, right? where do you have to run that? what directory? | 07:10 |
ltomasbo | also, I quickly check... you have a lot of leftovers from previous tempest run... it is possible that some new runs fails due to lack of quota | 07:11 |
digitalsimboja | tempest | 07:11 |
digitalsimboja | how do I clear previous runs? | 07:11 |
digitalsimboja | I wanted to do that in the night but couldn't | 07:12 |
ltomasbo | no magic I'm aware of... you will to manually remove the vms/networks/subnets... | 07:12 |
ltomasbo | well.. no VMs in your case | 07:12 |
ltomasbo | but namespaces named "kuryr-namespace-xxx" or pods on the defult namespace named "kuryr-pod-Xxx" | 07:13 |
ltomasbo | probably some services named "kuryr-service-xxx" | 07:14 |
gryf | digitalsimboja, you need to run stestr from tempest directory. | 07:18 |
digitalsimboja | ohh! | 07:19 |
digitalsimboja | I am just taking a look. Thanks | 07:19 |
digitalsimboja | https://paste.opendev.org/show/807748/ | 07:25 |
digitalsimboja | seems connection issues I guess? | 07:25 |
digitalsimboja | https://paste.opendev.org/show/807749/ | 07:29 |
digitalsimboja | I am having connection timeouts | 07:29 |
ltomasbo | digitalsimboja, did you cleaned up the env? | 07:32 |
ltomasbo | otherwise the extra amount of resources may be affecting you | 07:32 |
gryf | digitalsimboja, probably some devstack services has failed, due to no space left on device. maybe you need to start those again, although I suspect, that you'll need to unstack/stack again. | 07:34 |
ltomasbo | yeah, unstack and stack may be simpler | 07:35 |
digitalsimboja | cool | 07:39 |
digitalsimboja | https://paste.opendev.org/show/807750/ | 07:52 |
digitalsimboja | stacking hitting a block | 07:52 |
digitalsimboja | https://paste.opendev.org/show/807751/ | 08:04 |
ltomasbo | you need to check that is the problem (journalctl) | 08:10 |
opendevreview | Roman Dobosz proposed openstack/kuryr-kubernetes master: [WIP] Fix multinode gate, separate variations for OVN and OVS. https://review.opendev.org/c/openstack/kuryr-kubernetes/+/801618 | 09:43 |
opendevreview | Roman Dobosz proposed openstack/kuryr-kubernetes master: [WIP] Fix multinode gate, separate variations for OVN and OVS. https://review.opendev.org/c/openstack/kuryr-kubernetes/+/801618 | 11:08 |
digitalsimboja | @maysams, @ltomasbo: Just a quick one on the kuryr-tempest-plugin | 12:41 |
digitalsimboja | I understand I need to add the test on the config.py? | 12:42 |
digitalsimboja | The question is where do I define the test function? | 12:43 |
digitalsimboja | With regards to : | 12:44 |
digitalsimboja | https://paste.opendev.org/show/807762/ | 12:44 |
digitalsimboja | Say on this line : if not CONF.kuryr_kubernetes.test_sctp_services: | 12:45 |
digitalsimboja | raise cls.skipException("Service SCTP tests are not enabled") | 12:45 |
digitalsimboja | I want to find where test_sctp_services is defined? | 12:46 |
ltomasbo | those are flags for some test that should not run in all the configurations | 13:01 |
ltomasbo | for yours, probably you don't need to add any config function | 13:01 |
digitalsimboja | perfect | 13:04 |
ltomasbo | maybe just reuse some of the existing configs (such a service_tests_enabled) | 13:06 |
digitalsimboja | got it! but not defining a specific one per my test? | 13:14 |
ltomasbo | that can be decided later, I think (at least for now) it is not needed | 13:15 |
digitalsimboja | got it. Thanks | 13:15 |
digitalsimboja | @ltomasbo, @maysams: Just to clarify | 14:40 |
digitalsimboja | This is the approach I am taking: | 14:41 |
digitalsimboja | 1. Create as ervice | 14:41 |
digitalsimboja | 1. Create a service | 14:41 |
digitalsimboja | 2. Delete the associated loadbalancers on the service | 14:41 |
digitalsimboja | 3. Wait for the reconciliation timeout | 14:41 |
digitalsimboja | Now my question is on number 2 | 14:41 |
digitalsimboja | I guess that means I need to set the status to {}? | 14:42 |
digitalsimboja | Is that correct? | 14:42 |
gryf | status of what? | 14:42 |
digitalsimboja | the CRD | 14:42 |
gryf | well, no. | 14:43 |
ltomasbo | digitalsimboja, you need to remove the loadbalancer from openstack side | 14:43 |
digitalsimboja | https://paste.opendev.org/show/807769/ | 14:43 |
gryf | think of tempest tests as "real life", so it might happened that loadbalancer could be removed by mistake. | 14:43 |
ltomasbo | and ensure your reconciliation removes the status, to trigger the recreation of the loadbalancer | 14:43 |
gryf | so that you need to actually remove LB from octavia. | 14:44 |
gryf | as ltomasbo already said. | 14:44 |
ltomasbo | :) | 14:44 |
digitalsimboja | So I am removing LB from Octavia right on the tempest right and not setting the status of the CRD to 'empty'? | 14:44 |
gryf | yes. | 14:45 |
ltomasbo | the status of the CRD should be handle by the patch you already merged on kuryr-kubernetes | 14:45 |
ltomasbo | tempest side is to test that what you merged actually works | 14:45 |
digitalsimboja | That means my above implementation is wrong | 14:45 |
digitalsimboja | :) | 14:46 |
ltomasbo | so, as gryf mention, think of tempest as the same tests you were manually doing for testing your kuryr-kubernetes reconciliation patch | 14:46 |
gryf | imagine you want to test if your code works - you'll just remove the lb and wait, to see if it works, right? | 14:46 |
digitalsimboja | sure @gryf | 14:46 |
gryf | so the tempest test is just "you" doing this as a code. | 14:46 |
ltomasbo | what you were doing to ensure your patch was working was removing the loadbalancer and wait for your code to detect it, and recreate it. So that is what tempest test need to do | 14:46 |
digitalsimboja | understood | 14:46 |
digitalsimboja | understood | 14:47 |
ltomasbo | gryf, seems we are aligned in our thinking! xD | 14:47 |
gryf | ltomasbo, ;) | 14:47 |
digitalsimboja | Yeah! You both are very much aligned | 14:47 |
digitalsimboja | Let me reconfirm: I still need to get the openstackSDK client similarly same way as per my patch on tempest? Correct? | 14:48 |
ltomasbo | digitalsimboja, check other tests on kuryr-tempest-plugin to see how those are done | 14:50 |
digitalsimboja | Got it! | 14:50 |
ltomasbo | they are both creating/checking OpenStack and Kubernetes resources | 14:51 |
simboja | @ltomasbo: seems i missed few messages you sent regarding this?^ | 16:09 |
simboja | they are both creating/checking OpenStack and Kubernetes resources | 16:09 |
ltomasbo | that was the last one | 16:17 |
simboja | cool Thanks | 16:37 |
digitalsimboja | @dulek, @gryf, @maysams: I am yet to fiend a way of calling octavia to delete the loadbalancers on tempest | 18:14 |
digitalsimboja | my research took me here | 18:14 |
digitalsimboja | https://docs.openstack.org/api-ref/network/v2/index.html?expanded=confirm-a-tag-detail#list-service-providers | 18:14 |
maysams | digitalsimboja: have you checked the tempest repository? | 18:14 |
gryf | btw, on irc you can skip the '@' before nickname as it here does nothing ;) | 18:14 |
digitalsimboja | ohh | 18:15 |
digitalsimboja | tempest repo on github? | 18:15 |
digitalsimboja | checking the repo | 18:20 |
digitalsimboja | i tried pr = cls.os_admin.service_providers_client.list_service_providers() | 18:23 |
*** pmannidi|AFK is now known as pmannidi | 23:08 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!