[
https://issues.apache.org/jira/browse/HBASE-19266?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16264204#comment-16264204
]
Chia-Ping Tsai commented on HBASE-19266:
----------------------------------------
bq. The above occurred multiple times during a run with EAGER policy. It would
not cause the region server to shutdown.
The error I mentioned is shown below.
{quote}
2017-11-22 17:24:53,493 FATAL [RS_CLOSE_REGION-asf911:51862-0]
regionserver.HRegionServer(2376): ***** ABORTING region server
asf911.gq1.ygridcore.net,51862,1511371099686: Assertion failed while closing
store TestAcidGuarantees,,1511371469395.68d2d57885546cb47faa7361cbbbcca7. A.
flushableSize expected=0, actual= dataSize=2290 , heapSize=7008. Current
memstoreSize=-34350. Maybe a coprocessor operation failed and left the memstore
in a partially updated state. *****
2017-11-22 17:24:53,493 FATAL [RS_CLOSE_REGION-asf911:51862-0]
regionserver.HRegionServer(2382): RegionServer abort: loaded coprocessors are:
[org.apache.hadoop.hbase.coprocessor.MultiRowMutationEndpoint]
2017-11-22 17:24:53,557 INFO [RS_CLOSE_REGION-asf911:51862-0]
regionserver.HRegionServer(2386): Dump of metrics as JSON on abort: {
{quote}
The detail log is
[here|https://builds.apache.org/job/HBASE-Flaky-Tests/23491/artifact/hbase-server/target/surefire-reports/org.apache.hadoop.hbase.TestAcidGuarantees-output.txt/*view*/].
The code about aborting the rs is shown below.
{code:title=HRegion.class}
for (HStore store : stores.values()) {
MemStoreSize flushableSize = store.getFlushableSize();
if (!(abort || flushableSize.getDataSize() == 0 ||
writestate.readOnly)) {
if (getRegionServerServices() != null) {
getRegionServerServices().abort("Assertion failed while closing
store "
+ getRegionInfo().getRegionNameAsString() + " " + store
+ ". flushableSize expected=0, actual= " + flushableSize
+ ". Current memstoreSize=" + getMemStoreSize() + ". Maybe a
coprocessor "
+ "operation failed and left the memstore in a partially
updated state.", null);
}
}
{code}
bq. TestAcidGuaranteesWithBasicPoli didn't finish in the QA run.
I have ran TestAcidGuaranteesWithBasicPolicy 50 times, and I don't encounter
the hang...Let me try the v2 patch (asap)
{quote}
w.r.t. 'memstoreSize to a negative value' error, the first occurrence is in
TestAcidGuarantees#testMixedAtomicity
However, if I run the subtest alone, it passes with EAGER policy.
{quote}
It seems the error happens by chance. I will dig in after committing this issue.
> TestAcidGuarantees should cover adaptive in-memory compaction
> -------------------------------------------------------------
>
> Key: HBASE-19266
> URL: https://issues.apache.org/jira/browse/HBASE-19266
> Project: HBase
> Issue Type: Test
> Reporter: Ted Yu
> Assignee: Chia-Ping Tsai
> Priority: Minor
> Attachments: HBASE-19266.v0.patch
>
>
> Currently TestAcidGuarantees populates 3 policies of (in-memory) compaction.
> Adaptive in-memory compaction is new and should be added as 4th compaction
> policy.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
