Siyao Meng created HDDS-17192:
---------------------------------

             Summary: Add snapshot coverage to upgrade acceptance test
                 Key: HDDS-17192
                 URL: https://issues.apache.org/jira/browse/HDDS-17192
             Project: Apache Ozone
          Issue Type: Test
            Reporter: Siyao Meng


HDDS-17151 found that an OM that ran a snapshot diff before a downgrade could 
not start with the old version. The upgrade acceptance test did not catch it, 
for two reasons:

* The common non-rolling upgrade callback only writes and reads keys. The 
snapshot checks in the 1.4.0 callback do not run for the 2.2.1 to current 
upgrade.
* After each restart the test only waits for an OM leader, and {{ozone admin om 
roles}} reports peers from config, so a single OM that failed to start goes 
unnoticed.

Proposal: in each phase of the non-rolling upgrade, create a snapshot, read 
keys through {{.snapshot}} paths of snapshots taken in earlier phases, and diff 
against the previous snapshot (including in the pre-finalized phase, so the new 
version writes db.snapdiff before the downgrade). After each restart, wait for 
the RPC port of every OM.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to