[
https://issues.apache.org/jira/browse/HBASE-10988?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13974341#comment-13974341
]
Lars Hofhansl commented on HBASE-10988:
---------------------------------------
Some tests are still failing.
{code}
org.apache.hadoop.hbase.thrift.TestThriftServerCmdLine.testRunThriftServer[10]
org.apache.hadoop.hbase.thrift.TestThriftServerCmdLine.testRunThriftServer[11]
org.apache.hadoop.hbase.thrift.TestThriftServerCmdLine.testRunThriftServer[13]
org.apache.hadoop.hbase.thrift.TestThriftServerCmdLine.testRunThriftServer[14]
org.apache.hadoop.hbase.thrift.TestThriftServerCmdLine.testRunThriftServer[16]
org.apache.hadoop.hbase.thrift.TestThriftServerCmdLine.testRunThriftServer[17]
org.apache.hadoop.hbase.thrift.TestThriftServerCmdLine.testRunThriftServer[18]
{code}
This suggests that 10s might be too short as a wait time here.
I'd like to do an addendum (or a new issue) which increases the wait from 10s
to 20s. Any objections to that?
> Properly wait for server in TestThriftServerCmdLine
> ---------------------------------------------------
>
> Key: HBASE-10988
> URL: https://issues.apache.org/jira/browse/HBASE-10988
> Project: HBase
> Issue Type: Bug
> Components: test
> Reporter: Lars Hofhansl
> Assignee: Lars Hofhansl
> Priority: Minor
> Fix For: 0.99.0, 0.94.19, 0.98.2, 0.96.3
>
> Attachments: 10988-0.94.txt, 10988-trunk.txt
>
>
> In 0.94 I find:
> {{Threads.sleepWithoutInterrupt(2000)}}
> In trunk I see:
> {code}
> while ( thriftServer.serverRunner == null ||
> thriftServer.serverRunner.tserver == null ){
> Thread.sleep(1);
> }
> {code}
> Both aren't good.
> The 0.94 version will fail if the server does not come up within 2s. The
> trunk version (1) might wait forever and cause a long timeout for the test
> and (2) wait quite busily with only 1ms of sleeping.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
