Bug 1749557
| Summary: | panic in image signature controller | ||
|---|---|---|---|
| Product: | OpenShift Container Platform | Reporter: | Ben Parees <bparees> |
| Component: | ImageStreams | Assignee: | Gabe Montero <gmontero> |
| Status: | CLOSED ERRATA | QA Contact: | XiuJuan Wang <xiuwang> |
| Severity: | unspecified | Docs Contact: | |
| Priority: | unspecified | ||
| Version: | 4.2.0 | CC: | adam.kaplan, aos-bugs, gmontero, jokerman, wzheng |
| Target Milestone: | --- | ||
| Target Release: | 4.2.0 | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | If docs needed, set a value | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2019-10-16 06:40:35 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
|
Description
Ben Parees
2019-09-05 20:36:48 UTC
Ben / Adam (with Oleg out on PTO) Traversing the stack, the initial call at https://github.com/openshift/openshift-controller-manager/blob/master/pkg/image/controller/signature/container_image_downloader.go#L39 passes a nil context that is propagated down the containers/image methods and flagged by /usr/local/go/src/net/http/request.go:350 You would think this panic would show up frequently if this code path is called with any frequency. Seems like the context creation a few lines down should be moved up and passed into reference.NewImageSource(nil, nil) Thoughts? Though maybe the recent containers/image bump has inserted more stringent checks. i think imagesignature import is not exercised regularly, but i would also guess that it's being exercised here by a specific test so you'd think the panic would show up every time the test is run if nothing else. I wonder if this was introduced by a recent dep bump. Anyway in principal i agree w/ the suggested solution... though it might be worth checking how the code used to run to ensure the dep bump didn't introduce a remote call that we weren't even making previously, somehow. heh. comment collision. i think we're on the same page, gabe. Yep it did introduce remote calls .... if you look at the old https://github.com/openshift/openshift-controller-manager/blob/f65d698f67c6cc0f38b1977f8cb2a8e5d519217f/vendor/github.com/containers/image/docker/docker_image_src.go and the newImageSource method vs. what is there now ... where there is a testImageSource.ensureManifestIsLoaded(ctx) So I'll get a PR up with the suggested solution But yeah, the bump PRs ran the e2e-aws test suite ... let's keep an eye on it when the PR is up. [1]. Don't meet the imageimport error in last several e2e-aws ci job history. [2].Also when I try to refer to https://bugzilla.redhat.com/show_bug.cgi?id=1722568#c10 to get the image imagesignatures, just found there is no way to change openshift-controller-manager operator stand-by. $oc describe co openshift-controller-manager Status: Conditions: Last Transition Time: 2019-09-09T01:03:44Z Message: WorkloadDegraded: the controller manager spec was set to Unmanaged state, but that is unsupported, and has no effect on this condition Reason: AsExpected Status: False Type: Degraded So I will verify this bug refer to [1], and will figure out how to make [2] works. Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2019:2922 |