-
Notifications
You must be signed in to change notification settings - Fork 2.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Permanent build failure: Unpacking kubernetes release ... gzip: stdin: not in gzip format #990
Comments
Not sure why |
https://github.com/kubernetes/kubernetes/blob/master/cluster/get-kube.sh#L206 If you run curl without the -f flag, it doesn't exit with nonzero if it talked to the server but got an error response code. |
Right, that's what I thought I wanted, but I got confused by the manpage. $ curl -f https://storage.googleapis.com/kubernetes-federation-release/ci/v1.5.0-alpha.2.540+ccaba2ccd2231d/kubernetes.tar.gz -o kubernetes.tar.gz
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
0 0 0 0 0 0 0 0 --:--:-- --:--:-- --:--:-- 0
curl: (22) The requested URL returned error: 403 Forbidden I'll update |
Not I. Perhaps @colhom or @madhusudancs ? |
So, interestingly enough fetching a tarball from a very recent successful build works fine- which leads me to believe it's not the ACLs on the bucket. Investigating further. |
@ixdy I don't think ACLs changed recently. Also, ACLs on the bucket looks very similar to the |
I think the problem here is GCS's eventual consistency on bucket listing. From https://cloud.google.com/storage/docs/consistency
If you look at federation-build 3285, you'll see the lines:
It uploads the artifacts, then lists them and changes the ACLs. If we're unlucky, that second listing will miss some of the artifacts. The proper fix here is to upload the files with the public_read ACL in the first place. We could also sleep for some period of time to reduce the chances of getting bitten by eventual consistency. |
This fixes kubernetes/test-infra#990 where a recursive acl change immediately after an upload might miss objects, since GCS bucket listing is eventually consistent.
@rmmh Rather than sleep for some arbitrary period of time, please poll until the desired consistency is reached. |
This fixes kubernetes/test-infra#990 where a recursive acl change immediately after an upload might miss objects, since GCS bucket listing is eventually consistent.
Automatic merge from submit-queue Use curl -f in cluster/get-kube.sh **What this PR does / why we need it**: Make `curl` explicitly fail if there are HTTP errors. This makes things more obvious than tar mysteriously failing. x-ref kubernetes/test-infra#990 (I also added -L so HTTP redirects work, in case we eventually want to use them.)
We've been seeing failures like the following in the federation e2e tests on Jenkins since around midnight last night (and on occasion yesterday too). Anyone know anything about this?
Unpacking kubernetes release v1.5.0-alpha.2.540+ccaba2ccd2231d
gzip: stdin: not in gzip format
tar: Child returned status 1
tar: Error is not recoverable: exiting now
Recent examples in gubernator:
https://k8s-gubernator.appspot.com/build/kubernetes-jenkins/logs/kubernetes-e2e-gce-federation/5475
https://k8s-gubernator.appspot.com/build/kubernetes-jenkins/logs/kubernetes-e2e-gce-federation/5465
cc @kubernetes/test-infra-admins @kubernetes/sig-testing
Q
The text was updated successfully, but these errors were encountered: