[
https://issues.apache.org/jira/browse/HBASE-19898?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16345664#comment-16345664
]
Allen Wittenauer commented on HBASE-19898:
------------------------------------------
The mvn command never run. It looks like the job is hitting either the Yetus
resource limits or the global resource limits:
https://builds.apache.org/job/PreCommit-HBASE-Build/11260/artifact/patchprocess/coprocessors.txt
Add something like --proclimit=5000 to the command line, which is slightly less
than half of the max processes that the ASF Infra team has configured. Any more
than that and jobs will either randomly fail (best case) or cause the node to
fail (worst case, see HDFS-12711).
See INFRA-15685 where I'm trying to get it raised to something more reasonable.
> Canary should choose RegionStdOutSink automatically when write sniffing is
> specified
> ------------------------------------------------------------------------------------
>
> Key: HBASE-19898
> URL: https://issues.apache.org/jira/browse/HBASE-19898
> Project: HBase
> Issue Type: Bug
> Reporter: Ted Yu
> Assignee: Ted Yu
> Priority: Major
> Attachments: 19898.v1.txt
>
>
> Currently RegionServerStdOutSink is instantiated by default, even if user
> specifies -writeSniffing on the command line.
> Write sniffing would be ignored since Sink instance is of
> RegionServerStdOutSink class:
> {code}
> if (this.sink instanceof RegionServerStdOutSink || this.regionServerMode)
> {
> monitor =
> new RegionServerMonitor(connection, monitorTargets, this.useRegExp,
> (StdOutSink) this.sink, this.executor,
> this.regionServerAllRegions,
> this.treatFailureAsError);
> {code}
> RegionStdOutSink should be used for write sniffing.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
