QI, Jiale created HDFS-17950:
--------------------------------
Summary: Add FedBalance operational control switches
Key: HDFS-17950
URL: https://issues.apache.org/jira/browse/HDFS-17950
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: QI, Jiale
Large FedBalance jobs may need operator-controlled stop and resume points,
especially around the initial copy, incremental copy, and final cutover window.
This issue proposes adding optional operational control switches for
FedBalance. All new controls are disabled by default and do not change the
existing default behavior.
Proposed controls:
- Stop after the initial DistCp succeeds.
- Start a later job from the incremental DistCp stage after validating the
required prior-copy state.
- Stop instead of retrying forever when the diff is small enough but final-copy
conditions are blocked by open files.
- Gate leaving the incremental DistCp stage on a time-window sentinel path.
- Allow a sentinel path to make open files force-closeable.
These controls make long-running FedBalance migrations easier to operate while
preserving the current default behavior.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
