[07:46:53] accraze: o/ [07:47:04] I modified your test.sh script with [07:47:05] SERVICE_HOSTNAME="enwiki-goodfaith-predictor-default.kserve-test.example.com" [07:47:10] and the test works [07:48:08] I checked on the eqiad stack and the isvc command to get the URL seems not right anymore; [07:48:11] elukey@ml-serve-ctrl1001:~$ kubectl get isvc enwiki-goodfaith -n revscoring-editquality [07:48:14] NAME URL READY PREV LATEST PREVROLLEDOUTREVISION LATESTREADYREVISION AGE [07:48:17] enwiki-goodfaith 41d [07:49:24] the logs about ingress config are weird [07:50:20] and we have them in prod too /o\ [08:00:08] Unable to parse ingress config json: invalid character '\"' after object key:value pair [08:10:33] ok following the go code, it seems that i tried to fetch the ingress config contained in [08:10:36] kubectl get configmap inferenceservice-config -n kserve -oyaml [08:10:38] and indeed there is a typo [08:13:51] precisely https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/741844/ [08:13:58] added by yours truly [08:17:51] anyway, sandbox fixed :) [08:22:53] and prod should be fixed too now [08:23:20] kevinbazira: o/ I think that the sandbox should be ready for some tests if you want [11:28:04] * elukey lunch! [15:21:38] 10Lift-Wing, 10Machine-Learning-Team, 10Patch-For-Review: Create a LB service for inference.discovery.wmnet - https://phabricator.wikimedia.org/T289835 (10klausman) ` $ sudo confctl select 'cluster=ml_serve,service=kubesvc' set/pooled=yes:weight=1 The selector you chose has selected the following objects: {"... [16:34:34] 10Machine-Learning-Team, 10Patch-For-Review, 10Platform Team Initiatives (API Gateway), 10Platform Team Workboards (Platform Engineering Reliability): Proposal: add a per-service rate limit setting to API Gateway - https://phabricator.wikimedia.org/T295956 (10hnowlan) [17:19:54] klausman: o/ I tried to add the discovery service but we need to move inference's status away from lvs_setup (to something that is monitored and pages for example) [17:20:34] in case we can do it tomorrow or later on [17:20:48] I'd prefer tomorrow [17:20:50] I think that we can close the task for the codfw cluster [17:20:59] and keep open only the LVS one [17:21:03] sgtm [17:21:07] super [18:27:30] * elukey afk!