Chi-Hsuan Huang created HDDS-17102:
--------------------------------------
Summary: OM terminates with exit status 0 when Ratis shuts down
its state machine
Key: HDDS-17102
URL: https://issues.apache.org/jira/browse/HDDS-17102
Project: Apache Ozone
Issue Type: Bug
Components: OM
Reporter: Chi-Hsuan Huang
When Ratis closes the OM state machine on its own,
{{OzoneManagerStateMachine.close()}} calls {{OzoneManager.shutDown()}}, which
terminates with {{ExitUtils.terminate(0, ...)}}. A supervisor that restarts
only on failure treats this as a clean stop, and the termination is logged at
INFO. HDDS-16838 shows one such case: an NPE on the Ratis apply thread stops
the OM with status 0.
Status 0 came from HDDS-6930, when a normal stop also reached this path. A
normal stop no longer gets here: {{OzoneManager.stop()}} sets the state to
STOPPED before stopping Ratis, and {{close()}} skips {{shutDown()}} when OM is
already stopped. Other fatal OM paths, such as {{terminateOM()}}, already use
status 1.
One expected shutdown also goes through this path. Ratis shuts down a peer that
is removed from the ring, so an OM removed by {{ozone admin om decommission}}
likely ends here too (verified for SCM in HDDS-16922, not yet for OM). Changing
the status to 1 should keep status 0 for that case.
The fix should follow whatever HDDS-16922 settles on for SCM.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
