-
Bug
-
Resolution: Done-Errata
-
Undefined
-
None
-
4.13.0
-
Important
-
No
-
CNF RAN Sprint 233, CNF RAN Sprint 234, CNF RAN Sprint 235, CNF RAN Sprint 236, CNF RAN Sprint 237, CNF RAN Sprint 238
-
6
-
Rejected
-
False
-
-
Description of problem:
ptp operator socket management need rework since a few test case fails due to cleaning up the file before other processes are terminated.
linux ptp daemon has two differen Unix domain socket dependecy. 1. ptp4l/phc2sys and probably ts2phc uses same Unix domain socket used for local PTP management. 2. sidecar listens to event socket and each process in linux ptp daemon writes to it. In current implmentation when any process reecives SIGTERM, the cleanup process kills commonly used ptp management socket. This disturbs otther running process.For an example if phc2sys is stopped, it cleans up socket file, this puts ptp4l in error state and side car stopes recieving any messages.
Version-Release number of selected component (if applicable):
How reproducible:
everytime, on delete of ptpconfig , the process that terminates one ptp process puts other process in error state and side car is not able to recieve any updates after first process is killed. This puts process status event metrics with wrong values .
Steps to Reproduce:
1.apply ptp config 2.delete ptp config 3.watch logs , you will see socket error.
Actual results:
Expected results:
Additional info:
- blocks
-
OCPBUGS-12256 ptp operator socket management need rework since a few test case fails due to cleaning up the file before other processes are terminated.
- Closed
- is cloned by
-
OCPBUGS-12256 ptp operator socket management need rework since a few test case fails due to cleaning up the file before other processes are terminated.
- Closed
- is depended on by
-
OCPBUGS-12204 Backport of OCPBUGS-8112 to OCP 4.12
- Closed
- links to
-
RHEA-2023:5006 rpm