xxishu commented on PR #5641:
URL: https://github.com/apache/hbase/pull/5641#issuecomment-1902168800
Related to this issue, there are two independent options that can be
combined for configuration: whether readonly mode is enabled and whether
replicas are enabled.
There are four possible combinations:
1. Table with readonly mode enabled and replicas enabled.
2. Table with readonly mode disabled and replicas enabled.
3. Table with readonly mode enabled and replicas disabled.
4. Table with readonly mode disabled and replicas disabled.
If the table is both in readonly mode and replicas are disabled, the
StorefileRefresherChore will not be triggered, so it is not discussed further.
There are two implementation approaches:
1. Check if it is the primary region; if it is, skip the process.
2. Check if the table is in readonly mode; if it is, skip the process.
The key difference between these approaches is when the table is in readonly
mode and replicas are enabled, whether the Chore needs to be triggered. If the
Chore needs to be triggered, even after the table has been set to readonly, it
will continue checking replica regions. If the Chore interval is set too short,
it may result in a significant refresh of HDFS audit logs.
Additionally, as mentioned in
[propagating_writes_to_region_replicas](https://hbase.apache.org/book.html#_propagating_writes_to_region_replicas),
triggering the Chore is not necessary after the table is set to readonly.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
