[
https://issues.apache.org/jira/browse/HBASE-17482?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15829859#comment-15829859
]
Yu Li commented on HBASE-17482:
-------------------------------
bq. It looks like when HBASE-16698 was ported to master branch, some code is
lost.
Actually HBASE-16698 was reverted once in master branch, and before it's
committed again, when I prepared the patch for branch-1, I noticed that the
change HBASE-16768 made cannot collaborate well with HBASE-16698, so I fixed it
with the line you posted. However, the last commit of HBASE-16698 for master
branch is not done by me so possibly the change got missed. (The most strange
thing is that UT fails w/o my addendum fix in branch-1 but not in master, maybe
our test coverage is reduced somehow... [~stack] FYI sir.)
You could see the below line introduced by HBASE-16768 if look into patch there
or commit history.
{code}
boolean updateSeqId = replay || batchOp.getMutation(i).getDurability() ==
Durability.SKIP_WAL;
{code}
> mvcc mechanism fails when using mvccPreAssign
> ---------------------------------------------
>
> Key: HBASE-17482
> URL: https://issues.apache.org/jira/browse/HBASE-17482
> Project: HBase
> Issue Type: Bug
> Affects Versions: 2.0..
> Reporter: Allan Yang
> Assignee: Allan Yang
> Priority: Critical
> Fix For: 2.0.0
>
> Attachments: HBASE-17482.patch, HBASE-17482.v2.patch,
> HBASE-17482.v3.patch
>
>
> If mvccPreAssign and ASYNC_WAL is used, then cells may have been commited to
> memstore before append thread can stamp seqid to them. The unit test shows
> everything.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
