[
https://issues.apache.org/jira/browse/HBASE-23286?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16973866#comment-16973866
]
Duo Zhang commented on HBASE-23286:
-----------------------------------
I think [~stack] wanted to ask why the compaction marker matters.
The point here is that, before HBASE-20724 we need to retain compaction markers
in the recovered.edits, because we need to finish a compaction if the RS
crashed after write the compaction marker. You can see HBASE-2231 for more
details. This means, even if we write all the data to to HFiles directly, we
still need to write the compaction markers to recovered.edits, as we need these
markers when initialize a region.
And after HBASE-20724, we do not need to write compaction markers to
recovered.edits any more, as now we rely on the metadata in HFile directly, so
now we just need to write the data to hfiles, which will greatly simplify the
split log logic.
> Improve MTTR: Split WAL to HFile
> --------------------------------
>
> Key: HBASE-23286
> URL: https://issues.apache.org/jira/browse/HBASE-23286
> Project: HBase
> Issue Type: Improvement
> Components: MTTR
> Reporter: Guanghao Zhang
> Assignee: Guanghao Zhang
> Priority: Major
>
> After HBASE-20724, the compaction event marker is not used anymore when
> failover. So our new proposal is split WAL to HFile to imporve MTTR. It has 3
> steps:
> # Read WAL and write HFile to region’s column family’s recovered.hfiles
> directory.
> # Open region.
> # Bulkload the recovered.hfiles for every column family.
> The design doc was attathed by a google doc. Any suggestions are welcomed.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
