Shangshu Qian created HBASE-29600:
-------------------------------------
Summary: Potential Feedback Loop in Write Ahead Log (WAL) Handling
Key: HBASE-29600
URL: https://issues.apache.org/jira/browse/HBASE-29600
Project: HBase
Issue Type: Bug
Affects Versions: 2.6.0
Reporter: Shangshu Qian
We find a potential feedback loop in the WAL handling that can cause
unconstrained retries.
# Delay in the ReplicaitonSourceShipper, which handles the replication of the
WAL entries.
# The delay causes some IO-related error during the parsing process, leading
to premature EOF and subsequent retries. Retry logic is documented in
HBASE-15984.
# The retry adds more load to the node, causing more WAL entries to be retried.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
