Tanuj Khurana created PHOENIX-7845:
--------------------------------------
Summary: ReplicationLogGroup initialization resilience to standby
cluster unavailability
Key: PHOENIX-7845
URL: https://issues.apache.org/jira/browse/PHOENIX-7845
Project: Phoenix
Issue Type: Sub-task
Reporter: Tanuj Khurana
Assignee: Tanuj Khurana
When a RegionServer starts and the HAGroupStore indicates SYNC mode, but the
standby namenode is unavailable, ReplicationLogGroup.init() fails completely
at {_}createPeerShardManager(){_}. The existing runtime fallback logic which
gracefully transitions SYNC → STORE_AND_FORWARD on failure — is never reached
because it lives downstream of the peer shard manager creation.
The fix removes eager peer shard manager creation from init() and instead
creates it on demand with a per-mode-session lifetime. This allows the existing
updateModeOnFailure() path to handle degradation naturally.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
