Loading...

XML

Word

Printable

Type: Bug
Resolution: Done
Priority: Major
Fix Version/s: netobserv-1.6.1
Affects Version/s: netobserv-1.4.2, openshift-4.14.z
Component/s: eBPF
Labels:

Work Type:
Quality / Stability / Reliability
Blocked:
False
Blocked Reason:

Hide

None

Show
None
Story Points:
None
Severity:
None

Target Version:
None
Release Blocker:
None
Sprint:
NetObserv - Sprint 255, NetObserv - Sprint 256

SFDC Cases Counter:
SFDC Cases Open:
SFDC Cases Links:

Release Note Status:
None
Release Note Type:
None
Release Note Text:

Hide
Previously, the eBPF agent was unable to clean up tc flows installed before an ungraceful crash (e.g., due to a SIGTERM signal). This led to multiple tc flow filters with the same name being created without removing the older ones. With this fix, we ensure that all previously installed tc flows are cleaned up when the agent starts, before installing new ones.

Show
Previously, the eBPF agent was unable to clean up tc flows installed before an ungraceful crash (e.g., due to a SIGTERM signal). This led to multiple tc flow filters with the same name being created without removing the older ones. With this fix, we ensure that all previously installed tc flows are cleaned up when the agent starts, before installing new ones.

Description of problem:
In OCP 4.14.23 with nodes having cgroup v1 + network-observability-operator.v1.4.2

it is seen that netobserv-ebpf-agent pods are not releasing the memory after getting oomkilled , That causes the operator pod to get kill again with no task in the cgroup and saying no killable process found. This happens in loop.

Steps to Reproduce: