Bug 1946593
Summary: | Daemonset hostPort UDP conntrack entries are not updated when pods are recreated | ||||||
---|---|---|---|---|---|---|---|
Product: | OpenShift Container Platform | Reporter: | Andre Costa <andcosta> | ||||
Component: | Node | Assignee: | Ryan Phillips <rphillips> | ||||
Node sub component: | Kubelet | QA Contact: | Sunil Choudhary <schoudha> | ||||
Status: | CLOSED ERRATA | Docs Contact: | |||||
Severity: | high | ||||||
Priority: | high | CC: | aconstan, anbhat, aojeagar, aos-bugs, awallenb, bjarolim, ctauchen, jtanenba, mtapsonj, openshift-bugs-escalate, pehunt, rcarrier, schoudha, scuppett, vlaad, wzheng, zzhao | ||||
Version: | 3.11.0 | ||||||
Target Milestone: | --- | ||||||
Target Release: | 3.11.z | ||||||
Hardware: | x86_64 | ||||||
OS: | Linux | ||||||
Whiteboard: | |||||||
Fixed In Version: | Doc Type: | Bug Fix | |||||
Doc Text: |
Previously, for pods that used the `hostPort` definition to expose UDP ports to the host, the kubelet did not remove stale routing entries when a pod was deleted. As a result, those ports became unreachable when the pod was restarted. With this update, stale routing entries are removed, and the exposed UDP ports are reachable when the pods are restarted.
|
Story Points: | --- | ||||
Clone Of: | Environment: | ||||||
Last Closed: | 2021-08-04 11:18:17 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: | |||||||
Attachments: |
|
Description
Andre Costa
2021-04-06 13:21:15 UTC
This bug reported on 3.11.z and reproduces all the way up to 4.6.z. Setting source release to 3.11 and target to 4.8.0 to get a fix in our development branch and then clone/backport as far as needed/requested. Hi Everyone. Because RHCOS and toolbox have neither the conntrack-tools I was able to do that with this: https://access.redhat.com/articles/5929341 and did it solve the issue? I continue to be unable to reproduce, but I still believe the bug exists. @andcosta would you be able to use `rpm-ostree install` to install `conntrack-tools` and restart the node, and see if this issue persists? If adding conntrack fixes the issue, I'll work with coreOS team to get it in RHCOS It turns out that opernshift/dockershim hostport doesn't implement the conntrack deletion, CRIO does. Is this something we can test with steps on https://bugzilla.redhat.com/show_bug.cgi?id=1946593#c37 ? https://github.com/openshift/origin/pull/26206 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 (OpenShift Container Platform 3.11.487 bug fix and enhancement update), 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-2021:2928 |