Bug 2089392 - 120 node baremetal upgrade from 4.9.29 --> 4.10.13 crashloops on machine-approver
Summary: 120 node baremetal upgrade from 4.9.29 --> 4.10.13 crashloops on machine-app...
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: OpenShift Container Platform
Classification: Red Hat
Component: Networking
Version: 4.10
Hardware: x86_64
OS: Linux
medium
medium
Target Milestone: ---
: 4.11.0
Assignee: Jaime Caamaño Ruiz
QA Contact: Mike Fiedler
URL:
Whiteboard:
Depends On:
Blocks: 2099206
TreeView+ depends on / blocked
 
Reported: 2022-05-23 14:53 UTC by Dave Wilson
Modified: 2022-08-10 11:13 UTC (History)
6 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Cause: When adding a network policy, all other policies logging configuration was being updated unnecessarily keeping ovn-kuberenetes busy and not attending other policy related requests. Consequence: If there are a lot of network policies in the system, and a new one is added, there could be meaningful and noticeable latency on some concurrent or later network policy configuration changes being in effect. Fix: All network policies are no longer unnecessarily updated when adding a new policy. Result: Concurrent or later network policy configuration changes are applied in time.
Clone Of:
Environment:
Last Closed: 2022-08-10 11:13:36 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Github openshift ovn-kubernetes pull 1145 0 None open Bug 2089392: Update logging for specific policy when creating it 2022-06-15 13:57:45 UTC
Red Hat Product Errata RHSA-2022:5069 0 None None None 2022-08-10 11:13:54 UTC

Description Dave Wilson 2022-05-23 14:53:54 UTC
Description of problem:
120 node baremetal upgrade from 4.9.29 --> 4.10.13, with moderate pod scale,  crashloops on machine-approver for ~2hrs.

NAME                                READY   STATUS    RESTARTS         AGE
machine-approver-6c94df4c6c-fpdkn   2/2     Running   23 (5m27s ago)   124m
Mon May 23 00:52:12 UTC 2022
NAME                                READY   STATUS    RESTARTS   AGE
machine-approver-6c94df4c6c-7xbjj   2/2     Running   0          48s

F0522 23:50:17.693517       1 main.go:102] Can't create clients: failed to create client: Get "https://172.30.0.1:443/api?timeout=32s": dial tcp 172.30.0.1:443: i/o timeout


Version-Release number of selected component (if applicable):


How reproducible: 100%


Steps to Reproduce:
1. Deploy OCP 4.9.29 
2. scale BM env to 118 workers
3. instantiate pod scale with network policies ( https://github.com/cloud-bulldozer/kube-burner/tree/master/examples/workloads/kubelet-density-cni-networkpolicy)
4. Upgrade to 4.10.13

Actual results:
machine-approver pod crashloops for ~2hr but eventually reconciles but upgrade completion is delayed considerably 

Expected results:Successful upgrade in a reasonable time 


Additional info: must-gather available at: http://perf1.perf.lab.eng.bos.redhat.com/pub/dwilson/must-gather-machine-aprrover.tar.gz

Comment 2 Raul Sevilla 2022-05-26 11:37:44 UTC
Just to add more context, as Dave mentioned, we loaded the cluster with the kubelet-density-cni-networkpolicy workload.

Specifically 2500 iterations deploying:

- Creates namespace kubelet-density-cni-networkpolicy with the following objects
- 1 deny-all network policy in the
- 2500 webserver applications (nginx)
- 2500 services, each of them backing one of the previous applications
- 2500 network policies
- 2500 client applications. (cURL-ing the webserver service)

Comment 4 Dave Wilson 2022-06-13 14:48:53 UTC
PR that contains tentative fix (https://github.com/openshift/ovn-kubernetes/pull/1126) rectifies the issue and the upgrade complete in 1.5hrs, which is inline with expectations

Comment 11 Mike Fiedler 2022-06-27 13:09:33 UTC
Marking verified pre-merge per comment 4.

Comment 12 errata-xmlrpc 2022-08-10 11:13:36 UTC
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 (Important: OpenShift Container Platform 4.11.0 bug fix and security 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/RHSA-2022:5069


Note You need to log in before you can comment on or make changes to this bug.