istio-releases / daily-release Goto Github PK

default_proxy not set -> test passed
https://storage.googleapis.com/istio-prow/pull/istio-releases_daily-release/718/daily-e2e-dashboard/14/build-log.txt

default_proxy set -> test failed
https://storage.googleapis.com/istio-prow/pull/istio-releases_daily-release/718/daily-e2e-dashboard/13/build-log.txt

Other tests (e.g. daily_e2e_simple. daily_e2e_bookInfo, etc) with --default_proxy work fine however.

I0129 07:47:35.495] echosrv-deployment-1664016432-0wfgj    0/2       Init:ImagePullBackOff   0          3m
I0129 07:47:35.495] echosrv-deployment-1664016432-x13p5    0/2       Init:ImagePullBackOff   0          3m


I0129 07:45:24.647]  error: unable to upgrade connection: container not found ("echosrv")

Refer: https://k8s-gubernator.appspot.com/build/istio-prow/pull/istio-releases_daily-release/351/daily-e2e-rbac-auth-default/244/

The reasons behind is as following:

1. We build two istioctl in release, one has tag=docker.io baked while other has tag=gcr.io baked.
2. We don't push images to docker.io until final step of release and before, images only stay in gcr.io
3. In -default release-qulification tests, we use the istioctl baked with docker.io (which is the one customers should use after we publishing the release)

mixer/mixer_debug images seem to be messed up a little. Multiple daily tags pointing to the same image.

https://pantheon.corp.google.com/gcr/images/istio-release/GLOBAL/mixer?project=istio-release&organizationId=433637338589&gcrImageListsize=50

https://pantheon.corp.google.com/gcr/images/istio-release/GLOBAL/mixer_debug?project=istio-release&organizationId=433637338589&gcrImageListsize=50

istio/test-infra's config.yaml points to files in istio-releases/daily-release. Some of these files test in a single namespace but they should be testing cluster wide. This will allow us to remove istio-one-namespace-auth.yaml and istio-one-namespace.yaml from the release archive, as described in istio/istio#4859.

This issue is in conjunction with istio/test-infra#795.

The new helm istioctl script is failing on master.This might indicate an actual istioctl test failure but the logs are not very verbose and they doesn't point to a precise failure point.

Example failure https://storage.googleapis.com/istio-prow/pull/istio-releases_daily-release/1113/istioctl-tests/29/build-log.txt

Test:
https://github.com/istio-releases/daily-release/blob/master/prow/lib.sh
(BTW, we should rename it to be more descriptive instead of lib.sh)

Failure:
https://storage.googleapis.com/istio-prow/pull/istio-releases_daily-release/714/daily-e2e-bookinfoTests/7/build-log.txt

It is a good thing that the test successfully caught a broken change by istio/istio#6543, but there is no meaning error and it took us a while to figure it out. Also, with a test like this, how can we monitor on testgrid?