| *** gianpietro has quit IRC | 00:16 | |
| *** gianpietro has joined #openstack-kuryr | 00:21 | |
| *** gianpietro has quit IRC | 00:26 | |
| *** gianpietro has joined #openstack-kuryr | 00:27 | |
| *** maysamacedos has joined #openstack-kuryr | 00:30 | |
| *** gianpietro has quit IRC | 00:31 | |
| *** gianpietro has joined #openstack-kuryr | 00:38 | |
| *** gianpietro has quit IRC | 00:43 | |
| *** hongbin has joined #openstack-kuryr | 00:57 | |
| *** kiennt26 has joined #openstack-kuryr | 01:10 | |
| *** gianpietro has joined #openstack-kuryr | 01:49 | |
| *** yamamoto has joined #openstack-kuryr | 02:15 | |
| *** kiennt26 has quit IRC | 02:28 | |
| *** lihi has quit IRC | 02:36 | |
| *** lihi has joined #openstack-kuryr | 02:37 | |
| *** maysamacedos has quit IRC | 02:39 | |
| *** kiennt26 has joined #openstack-kuryr | 02:43 | |
| *** rh-jelabarre has quit IRC | 03:27 | |
| *** rh-jelabarre has joined #openstack-kuryr | 03:36 | |
| *** rh-jelabarre has quit IRC | 03:59 | |
| *** kiennt26 has quit IRC | 04:29 | |
| *** yboaron has joined #openstack-kuryr | 04:42 | |
| *** CrayZee has joined #openstack-kuryr | 04:57 | |
| *** gcheresh_ has joined #openstack-kuryr | 05:05 | |
| *** hongbin has quit IRC | 05:06 | |
| *** gcheresh_ has quit IRC | 05:10 | |
| *** gcheresh has joined #openstack-kuryr | 05:11 | |
| *** salv-orlando has joined #openstack-kuryr | 05:17 | |
| *** salv-orlando has quit IRC | 05:17 | |
| *** ispp has joined #openstack-kuryr | 06:12 | |
| *** isssp has quit IRC | 06:14 | |
| *** salv-orlando has joined #openstack-kuryr | 06:32 | |
| *** salv-orlando has quit IRC | 06:32 | |
| *** salv-orlando has joined #openstack-kuryr | 06:33 | |
| *** salv-orlando has quit IRC | 06:52 | |
| *** pcaruana has joined #openstack-kuryr | 06:54 | |
| *** yamamoto_ has joined #openstack-kuryr | 06:57 | |
| *** yamamoto has quit IRC | 07:00 | |
| *** pcaruana is now known as pcaruana|worksho | 07:03 | |
| *** gianpietro has quit IRC | 07:04 | |
| *** gianpietro has joined #openstack-kuryr | 07:05 | |
| *** gianpietro has quit IRC | 07:10 | |
| *** janki has joined #openstack-kuryr | 07:15 | |
| *** celebdor1 has joined #openstack-kuryr | 07:26 | |
| *** gianpietro has joined #openstack-kuryr | 07:29 | |
| *** pcaruana|worksho is now known as pcaruana | 07:50 | |
| *** garyloug has joined #openstack-kuryr | 08:04 | |
| *** yboaron has quit IRC | 08:20 | |
| *** salv-orlando has joined #openstack-kuryr | 08:49 | |
| *** salv-orlando has quit IRC | 08:55 | |
| *** gianpietro has quit IRC | 09:06 | |
| *** lxkong has quit IRC | 09:09 | |
| *** phuoc_ has joined #openstack-kuryr | 09:11 | |
| *** phuoc has quit IRC | 09:14 | |
| *** lxkong has joined #openstack-kuryr | 09:32 | |
| *** salv-orlando has joined #openstack-kuryr | 09:51 | |
| *** salv-orlando has quit IRC | 09:55 | |
| *** garyloug has quit IRC | 10:22 | |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Namespace deletion functionality for namespace_subnet driver https://review.openstack.org/562249 | 10:23 |
|---|---|---|
| *** garyloug has joined #openstack-kuryr | 10:26 | |
| celebdor1 | ltomasbo: ping | 10:34 |
| ltomasbo | celebdor1, pong! | 10:35 |
| celebdor1 | ltomasbo: IIRC when you do network delete for a network that has ports, it fails, right? | 10:35 |
| ltomasbo | celebdor1, yes | 10:36 |
| dulek | ltomasbo: Yay, it's merging! | 10:37 |
| celebdor1 | ok | 10:37 |
| ltomasbo | celebdor1, why you asked? | 10:38 |
| celebdor1 | ltomasbo: cause in your patch I don't see port removal | 10:40 |
| celebdor1 | and considering that the namespace deletion is a cascading op in k8s | 10:40 |
| celebdor1 | that makes namespace deletion a bit racy | 10:41 |
| celebdor1 | doesn't it? | 10:41 |
| ltomasbo | celebdor1, that was why I handled the case for pools in a follow up patch | 10:41 |
| celebdor1 | you rely on the pod deletion events to be processed before the namespace deletion event | 10:41 |
| celebdor1 | I am not talking about the pool case | 10:41 |
| celebdor1 | I'm talking about event races | 10:41 |
| ltomasbo | celebdor1, you mean the ports are being deleted but not yet deleted, and we triggered the network deletion, right? | 10:42 |
| celebdor1 | let's see if I get this right | 10:42 |
| celebdor1 | I have a namespace with 1000 pods | 10:42 |
| celebdor1 | when I send a DELETE to the namespace resource | 10:43 |
| celebdor1 | K8s will start doing DELETE events for each of those 1000 pods | 10:43 |
| celebdor1 | and then will send the delete event of the namespace | 10:43 |
| celebdor1 | I hope this is right | 10:43 |
| ltomasbo | yep, and kuryr (without pools) will start deleting the ports | 10:43 |
| celebdor1 | in case it is, what happens if we are slow deleting those pods and another green thread gets the namespace deletion | 10:44 |
| celebdor1 | it will fail to delete the network | 10:44 |
| celebdor1 | and fail to delete the crd | 10:44 |
| celebdor1 | right? | 10:44 |
| ltomasbo | just to be clear, this is without pools --> https://review.openstack.org/#/c/562249/ | 10:44 |
| ltomasbo | and if delete_network fails, the CRD will not be deleted, yes | 10:45 |
| ltomasbo | and on_delete will raise a NeutronClientException | 10:45 |
| ltomasbo | so, I guess there will be a retry, right? | 10:46 |
| ltomasbo | ohh, but perhaps I should split the remove_interface_router and delete network in 2 different try/except | 10:46 |
| celebdor1 | ltomasbo: my point is that probably we could raise a resource not ready | 10:47 |
| celebdor1 | so it is automatically retried | 10:47 |
| celebdor1 | and yes, in that case, the neutron deletion actions should have different error handling | 10:47 |
| ltomasbo | make sense | 10:47 |
| ltomasbo | celebdor1, can you remove the +W? | 10:47 |
| celebdor1 | I still merged the patch. I'm just asking for higher reliability | 10:47 |
| celebdor1 | in a follow up patch | 10:47 |
| celebdor1 | I don't need longer patch series nor complexity | 10:48 |
| celebdor1 | that deserves its own patch | 10:48 |
| ltomasbo | ok, then I'll add a following patch to address that | 10:48 |
| celebdor1 | thanks ltomasbo! | 10:50 |
| ltomasbo | thank you! | 10:50 |
| celebdor1 | ;-) | 10:50 |
| *** pliu has quit IRC | 10:50 | |
| *** pliu has joined #openstack-kuryr | 10:51 | |
| *** atoth has joined #openstack-kuryr | 11:16 | |
| celebdor1 | ltomasbo: regarding https://review.openstack.org/#/c/572118/1/kuryr_kubernetes/controller/drivers/vif_pool.py | 11:16 |
| celebdor1 | aren't the keys devices names and the values vif annotations? | 11:17 |
| celebdor1 | or are the port ovos? | 11:17 |
| celebdor1 | with the name you put it suggests the latter, but I thought it was the former | 11:17 |
| * ltomasbo looking | 11:18 | |
| ltomasbo | celebdor1, yes, keys are device names (like 'eth0') and the values the annotations | 11:18 |
| ltomasbo | and I need the port id present at the annotation, that is why I use annotatoins.values() | 11:19 |
| celebdor1 | ltomasbo: that part is clear | 11:19 |
| ltomasbo | well, it is more like --> {'eth0': vif-obj} | 11:19 |
| celebdor1 | I only meant that it should be | 11:19 |
| celebdor1 | for vif in annotations.values() | 11:19 |
| celebdor1 | instead of for port | 11:20 |
| celebdor1 | the port var name there is misleading | 11:20 |
| ltomasbo | ahh, yes, you are right! | 11:20 |
| ltomasbo | I'll change it right away | 11:20 |
| celebdor1 | thanks | 11:21 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Add ports pool clean up support to namespace deletion https://review.openstack.org/564148 | 11:23 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Ensure namespace delete is retry to avoid race https://review.openstack.org/572344 | 11:23 |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Fix precreated ports recovery after pod annotations change https://review.openstack.org/572118 | 11:28 |
| *** yamamoto_ has quit IRC | 11:29 | |
| ltomasbo | celebdor1, ^^ done! | 11:30 |
| celebdor1 | great | 11:31 |
| celebdor1 | irenab, dulek: please review https://review.openstack.org/#/c/572118/2 | 11:31 |
| dulek | celebdor1: Hm, I have a thought now… We probably need an convertion utility or compatibility code in case annotation is in old format. | 11:35 |
| dulek | In case of Kuryr upgrade from Queens. | 11:35 |
| *** yamamoto has joined #openstack-kuryr | 11:35 | |
| openstackgerrit | Luis Tomas Bolivar proposed openstack/kuryr-kubernetes master: Retry namespace deletion to mitigate cascading race https://review.openstack.org/572344 | 11:36 |
| celebdor1 | dulek: of course we do | 11:38 |
| celebdor1 | I'll think about it | 11:39 |
| celebdor1 | thanks for bringing it up! | 11:39 |
| dulek | celebdor1: Quick thought #2: since we don't support anyone running trunk yet, we should be able to undo some changes while it's still Rocky. | 11:45 |
| dulek | celebdor1: So I think the best way to approach this compatibility code would be to switch from putting blank dict to an dictionary o.vo. | 11:45 |
| dulek | celebdor1: That way you only fetch the annotation, check its type and if it's the old one, you pack it. | 11:46 |
| *** snapiri- has joined #openstack-kuryr | 11:48 | |
| *** CrayZee has quit IRC | 11:50 | |
| openstackgerrit | Merged openstack/kuryr-libnetwork master: modify the kuryr workflow document https://review.openstack.org/572085 | 11:52 |
| openstackgerrit | Merged openstack/kuryr-libnetwork master: [ci] Use zuul v3 native job for Rally https://review.openstack.org/568855 | 11:52 |
| *** snapiri- is now known as CrayZee | 11:53 | |
| *** CrayZee is now known as Guest91145 | 11:53 | |
| *** Guest91145 has quit IRC | 11:53 | |
| *** snapiri- has joined #openstack-kuryr | 11:54 | |
| *** snapiri- has quit IRC | 11:54 | |
| *** snapiri- has joined #openstack-kuryr | 11:55 | |
| *** snapiri- has quit IRC | 11:55 | |
| *** yamamoto has quit IRC | 11:58 | |
| *** rh-jelabarre has joined #openstack-kuryr | 11:59 | |
| celebdor1 | dulek: that makes sense | 12:07 |
| celebdor1 | a lot of it | 12:07 |
| celebdor1 | I was also wondering if it won't converge with your kuryrport work | 12:07 |
| celebdor1 | and then the annotation would just be a reference to a CRD | 12:07 |
| dulek | celebdor1: This could definitely help, but I think we should fix it in master first. | 12:09 |
| celebdor1 | agreed | 12:17 |
| *** maysamacedos has joined #openstack-kuryr | 12:19 | |
| *** yamamoto has joined #openstack-kuryr | 12:44 | |
| *** rh-jelabarre has quit IRC | 12:52 | |
| openstackgerrit | Merged openstack/kuryr-tempest-plugin master: Testing curl to the service of type LoadBalancer https://review.openstack.org/570734 | 12:54 |
| *** rh-jelabarre has joined #openstack-kuryr | 12:58 | |
| *** CrayZee has joined #openstack-kuryr | 13:15 | |
| *** celebdor1 has quit IRC | 13:27 | |
| *** celebdor1 has joined #openstack-kuryr | 13:30 | |
| *** yamamoto has quit IRC | 13:44 | |
| *** gcheresh has quit IRC | 14:19 | |
| *** hongbin has joined #openstack-kuryr | 14:21 | |
| *** maysamacedos has quit IRC | 14:29 | |
| *** maysamacedos has joined #openstack-kuryr | 14:30 | |
| *** yamamoto has joined #openstack-kuryr | 14:45 | |
| *** yamamoto has quit IRC | 14:51 | |
| *** yamamoto has joined #openstack-kuryr | 14:56 | |
| *** yamamoto has quit IRC | 15:01 | |
| *** celebdor1 has quit IRC | 15:01 | |
| *** pcaruana has quit IRC | 15:05 | |
| *** janki has quit IRC | 15:37 | |
| *** celebdor1 has joined #openstack-kuryr | 15:44 | |
| *** celebdor1 has quit IRC | 15:50 | |
| *** celebdor1 has joined #openstack-kuryr | 15:52 | |
| *** yamamoto has joined #openstack-kuryr | 15:58 | |
| *** yamamoto has quit IRC | 16:03 | |
| *** maysamacedos has quit IRC | 16:23 | |
| *** yboaron has joined #openstack-kuryr | 16:33 | |
| *** yamamoto has joined #openstack-kuryr | 16:50 | |
| *** yamamoto has quit IRC | 17:00 | |
| openstackgerrit | Merged openstack/kuryr-kubernetes master: Fix precreated ports recovery after pod annotations change https://review.openstack.org/572118 | 17:01 |
| *** celebdor1 has quit IRC | 17:03 | |
| *** maysamacedos has joined #openstack-kuryr | 17:10 | |
| *** isssp has joined #openstack-kuryr | 17:35 | |
| *** ispp has quit IRC | 17:40 | |
| *** yamamoto has joined #openstack-kuryr | 17:56 | |
| *** yamamoto has quit IRC | 18:02 | |
| *** CrayZee has quit IRC | 18:15 | |
| *** yamamoto has joined #openstack-kuryr | 18:58 | |
| *** yamamoto has quit IRC | 19:03 | |
| *** kzaitsev_pi has quit IRC | 19:29 | |
| *** kzaitsev_pi has joined #openstack-kuryr | 19:30 | |
| *** janki has joined #openstack-kuryr | 19:32 | |
| *** yamamoto has joined #openstack-kuryr | 19:59 | |
| *** yamamoto has quit IRC | 20:05 | |
| *** yboaron has quit IRC | 20:05 | |
| *** maysamacedos has quit IRC | 20:36 | |
| *** jerms has left #openstack-kuryr | 20:43 | |
| *** yamamoto has joined #openstack-kuryr | 21:01 | |
| *** yamamoto has quit IRC | 21:05 | |
| *** celebdor1 has joined #openstack-kuryr | 21:06 | |
| *** rh-jelabarre has quit IRC | 21:12 | |
| *** janki has quit IRC | 21:23 | |
| *** celebdor1 has quit IRC | 21:43 | |
| *** yamamoto has joined #openstack-kuryr | 22:02 | |
| *** atoth has quit IRC | 22:03 | |
| *** yamamoto has quit IRC | 22:06 | |
| *** yamamoto has joined #openstack-kuryr | 23:03 | |
| *** garyloug has quit IRC | 23:05 | |
| *** yamamoto has quit IRC | 23:07 | |
| *** threestrands has joined #openstack-kuryr | 23:08 | |
| *** hongbin has quit IRC | 23:27 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!