kubernetes unit tests fail with Panic messages stating "invalid type" #92070
Assignees
Labels
kind/bug
Categorizes issue or PR as related to a bug.
priority/important-soon
Must be staffed and worked on either currently, or very soon, ideally in time for the next release.
sig/api-machinery
Categorizes an issue or PR as relevant to SIG API Machinery.
Milestone
Comments
Rajalakshmi-Girish
added
the
kind/bug
k8s-ci-robot
added
the
needs-sig
Rajalakshmi-Girish
changed the title
Rajalakshmi-Girish
changed the title
/sig api-machinery |
k8s-ci-robot
added
sig/api-machinery
|
Sample stacktrace: |
|
ETCD has fixed this issue by etcd-io/etcd#12000. There still no semantic version for ETCD, I suggest updating the ETCD dependency with a pseudo version, can I? @liggitt |
|
/assign |
|
etcd was just updated in #92075, does that sha not include the fix? We use a sha in our go.mod because we cannot refer to etcd using a v3.4.x semver version, because it includes a go.mod file and did not rename its module to add a /v3 suffix. |
Unfortunately it does not include the fix. The sha in #92075 is Do you know how to create a
etcd already add /v3 suffix in its module. Should we update the import path in k/k accordingly and change it to a semantic version? |
|
Is etcd fixed in the 3.4.x branch? |
|
|
Can that fix be picked to 3.4.x compatibly? |
Sorry, I can't get your point, why you ask? And can you share a guideline about what we should do?
Let me give it a try. |
looks like it doesnot fix the below error: |
|
@Rajalakshmi-Girish Did you ever find a solution to the second bug above? I'm seeing the same after applying the original fix from etcd-io/etcd#12000. |
No I haven't. @liggitt Is there a solution? |
|
looks like etcd-io/etcd#12106 will pick that to etcd 3.4, then we'd need a 3.4.10 tag and a bump in k8s.io/kubernetes |
liggitt
added
the
priority/important-soon
Please report it to etcd similar to how the first bug was reported. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
logs_ci-unit-kubernetes_50_build-log.txt
What happened:
Below are the 27 unit tests that panic and FAIL while I run "make test"
Also the analysis says that the failure started occurring after the below PR merge:
#283d9f7
This commit updates github.com/golang/protobuf from v1.3.3 to v1.4.2.
The invalid Type failures that occur are from the newer implementation of protobuf that is added at "kubernetes/vendor/google.golang.org/protobuf" which did not exist prior to this commit.
Failure Message: panic: field etcdserverpb.Request.ID has invalid type: got uint64, want pointer
Failure message: panic: invalid Go type int for field k8s_io.kubernetes.vendor.go_etcd_io.etcd.etcdserver.etcdserverpb.loggablePutRequest.value_size
What you expected to happen:
The unit tests must pass without error message.
How to reproduce it (as minimally and precisely as possible):
eg: make test WHAT=k8s.io/kubernetes/pkg/registry/apps/deployment/storage/
Anything else we need to know?:
This failure consistently occurs when unit test is run through an automated Jenkins pipeline. But the failure is seen intermittently when run "make test" manually. Sometimes, easily reproducible when we do "make clean", "make clean_generated" followed by "make test"
Please see the consistent failure at https://k8s-testgrid.appspot.com/sig-node-ppc64le#unit-tests
Environment:
Architechture: linux/ppc64le
kubectl version):cat /etc/os-release): Ubuntu 18.04.4uname -a):@mkumatag
The text was updated successfully, but these errors were encountered: