Adar Dembo has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/12166 )

Change subject: KUDU-2652: deflake 
TsRecoveryITest.TestNoBlockIDReuseIfMissingBlocks
......................................................................


Patch Set 1:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/12166/1//COMMIT_MSG
Commit Message:

http://gerrit.cloudera.org:8080/#/c/12166/1//COMMIT_MSG@11
PS1, Line 11: This patch makes orphaned block
            : generation faster by making the workload update a single row.
Hmm, why is this obviously faster? Updating one row over and over means we'll 
flush one MRS (INSERT and n UPDATEs), then flushing that DRS's DMS over and 
over. I guess there'll be some delta compaction which will delete blocks, but 
is that de facto more likely than the merge compaction you'd expect to see on 
the tiny DRSes produced by the default write pattern?

I also wonder why this surfaced just now. I suspect Will's change to the merge 
compaction policy, but given the non-default values to flush_threshold_mb and 
flush_threshold_secs I would have expected the change to make merge compaction 
more likely not less.


http://gerrit.cloudera.org:8080/#/c/12166/1/src/kudu/integration-tests/ts_recovery-itest.cc
File src/kudu/integration-tests/ts_recovery-itest.cc:

http://gerrit.cloudera.org:8080/#/c/12166/1/src/kudu/integration-tests/ts_recovery-itest.cc@257
PS1, Line 257:     
write_workload->set_write_pattern(TestWorkload::UPDATE_ONE_ROW);
Should add a comment here explaining why we're using this.



--
To view, visit http://gerrit.cloudera.org:8080/12166
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I24a689702d29a744be8c113fedafec0307a90b1c
Gerrit-Change-Number: 12166
Gerrit-PatchSet: 1
Gerrit-Owner: Andrew Wong <[email protected]>
Gerrit-Reviewer: Adar Dembo <[email protected]>
Gerrit-Reviewer: Andrew Wong <[email protected]>
Gerrit-Reviewer: Kudu Jenkins (120)
Gerrit-Reviewer: Mike Percy <[email protected]>
Gerrit-Comment-Date: Sat, 05 Jan 2019 02:01:52 +0000
Gerrit-HasComments: Yes

Reply via email to