[
https://issues.apache.org/jira/browse/GEODE-6950?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17172695#comment-17172695
]
ASF subversion and git services commented on GEODE-6950:
--------------------------------------------------------
Commit 1a6e1927d88fe974d375097b3e10d357d9f2053a in geode's branch
refs/heads/develop from Bruce Schuchardt
[ https://gitbox.apache.org/repos/asf?p=geode.git;h=1a6e192 ]
GEODE-6950: hot loop in PrimaryHandler.processRequest() (#5422)
* GEODE-6950: Locator can't start if a lot of clients already started
I've implemented the suggested fix & added unit tests. The
"processRequest" test covers the fix. I added a couple of other tests to
make sure the other primary paths through
PrimaryHandler.processRequest() are covered.
* altered the test to have a deterministic clock
> Locator can't start if a lot of clients already started
> -------------------------------------------------------
>
> Key: GEODE-6950
> URL: https://issues.apache.org/jira/browse/GEODE-6950
> Project: Geode
> Issue Type: Bug
> Components: core, membership
> Affects Versions: 1.7.0, 1.8.0, 1.9.0, 1.10.0, 1.11.0, 1.12.0
> Reporter: Eugene Nedzvetsky
> Assignee: Bruce J Schuchardt
> Priority: Major
> Labels: pull-request-available
> Attachments: 1.log
>
>
> Locator can't start if a few hundred clients already started.
> Steps to reproduce:
> 1. Start Locator
> 2. Start 300 Geode clients
> 3. Stop Locator
> 4. Start Locator again
> Observe 100% CPU load and after some time Locator app crashes with timeout
> exceptions in the log.
> The problem is in the method
> org.apache.geode.distributed.internal.InternalLocator.PrimaryHandler#processRequest
> handlerMapping doesn't have handlers for LocatorListRequest and
> ClientConnectionRequest requests on Locator startup and in this case work
> code part with condition 'if(giveup == 0)'(InternalLocator:1185)
> Pause Thread.sleep(1000) works only on the first iteration and after that
> giveup>0 and CPU just spends resources on cycle execution without any pauses.
> Call Thread.sleep(1000) should be after if(giveup>0) condition block. It
> will be called on each iteration in this case.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
