You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@geode.apache.org by "Bill Burcham (Jira)" <ji...@apache.org> on 2020/01/03 23:04:00 UTC

[jira] [Commented] (GEODE-7648) Locator will not start if other locator hostnames are not resolvable

    [ https://issues.apache.org/jira/browse/GEODE-7648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17007822#comment-17007822 ] 

Bill Burcham commented on GEODE-7648:
-------------------------------------

Before GEODE-7167 was fixed, calling {{GMSUtil.parseLocators()}} with a non-null {{bindAddress}} that was not resolveable would cause an NPE.

After that fix, we no longer see throw NPE—instead we throw {{MembershipConfigurationException}} in that case.

But also as part of that "fix" we introduced a scenario where, if the {{bindAddress}} is null, rather than eliding the resolution check, we do the check anyway—and throw the {{MembershipConfigurationException}} if resolution fails.

This new behavior represents a regression. The fix is to elide the hostname resolution entirely, if {{bindAddress}} is null.

> Locator will not start if other locator hostnames are not resolvable
> --------------------------------------------------------------------
>
>                 Key: GEODE-7648
>                 URL: https://issues.apache.org/jira/browse/GEODE-7648
>             Project: Geode
>          Issue Type: Bug
>          Components: locator, membership
>    Affects Versions: 1.11.0, 1.12.0
>            Reporter: Jacob Barrett
>            Assignee: Bill Burcham
>            Priority: Major
>
> Starting a locator with a list of other locators will fail if any names in that list are not resolvable. This is a common occurrence in a Kubernetes where DNS entries for a container are not created until the container is created. Starting two locators will result in the first one failing until the second one starts.
> {noformat}
> java -server -classpath /geode/lib/geode-dependencies.jar -Djava.awt.headless=true -Dsun.rmi.dgc.server.gcInterval=9223372036854775806 -Dgemfire.log-file= -Dlog4j.configurationFile=/geode/config/log4j.xml -Dgemfire.enable-cluster-configuration=true -Dgemfire.launcher.registerSignalHandlers=true '-Dgemfire.locators=geodecluster-sample-locator-0.geodecluster-sample-locator[10334],geodecluster-sample-locator-1.geodecluster-sample-locator[10334]' -Dgemfire.enable-experimental-cluster-management-service=true -Dgemfire.load-cluster-configuration-from-dir=false org.apache.geode.distributed.LocatorLauncher start geodecluster-sample-locator-0 --port=10334
> [warn 2020/01/03 19:02:53.524 GMT <main> tid=0x1] Unknown locator host: geodecluster-sample-locator-1.geodecluster-sample-locator
> [warn 2020/01/03 19:02:53.532 GMT <main> tid=0x1] Unknown locator host: geodecluster-sample-locator-1.geodecluster-sample-locator
> [info 2020/01/03 19:02:53.860 GMT <main> tid=0x1] Starting peer location for Distribution Locator on geodecluster-sample-locator-0.geodecluster-sample-locator.default.svc.cluster.local[10334]
> [info 2020/01/03 19:02:53.868 GMT <main> tid=0x1] locator is exiting due to an exception
> org.apache.geode.GemFireConfigException: This process is attempting to use a locator at an unknown address or FQDN: geodecluster-sample-locator-1.geodecluster-sample-locator
>     at org.apache.geode.distributed.internal.membership.gms.GMSUtil.parseLocators(GMSUtil.java:139)
>     at org.apache.geode.distributed.internal.membership.gms.locator.GMSLocator.<init>(GMSLocator.java:106)
>     at org.apache.geode.distributed.internal.membership.adapter.GMSLocatorAdapter.<init>(GMSLocatorAdapter.java:50)
>     at org.apache.geode.distributed.internal.membership.NetLocatorFactory.newLocatorHandler(NetLocatorFactory.java:32)
>     at org.apache.geode.distributed.internal.InternalLocator.startPeerLocation(InternalLocator.java:612)
>     at org.apache.geode.distributed.internal.InternalLocator.startLocator(InternalLocator.java:373)
>     at org.apache.geode.distributed.LocatorLauncher.start(LocatorLauncher.java:681)
>     at org.apache.geode.distributed.LocatorLauncher.run(LocatorLauncher.java:588)
>     at org.apache.geode.distributed.LocatorLauncher.main(LocatorLauncher.java:210)
> Exception in thread "main" org.apache.geode.GemFireConfigException: This process is attempting to use a locator at an unknown address or FQDN: geodecluster-sample-locator-1.geodecluster-sample-locator
>     at org.apache.geode.distributed.internal.membership.gms.GMSUtil.parseLocators(GMSUtil.java:139)
>     at org.apache.geode.distributed.internal.membership.gms.locator.GMSLocator.<init>(GMSLocator.java:106)
>     at org.apache.geode.distributed.internal.membership.adapter.GMSLocatorAdapter.<init>(GMSLocatorAdapter.java:50)
>     at org.apache.geode.distributed.internal.membership.NetLocatorFactory.newLocatorHandler(NetLocatorFactory.java:32)
>     at org.apache.geode.distributed.internal.InternalLocator.startPeerLocation(InternalLocator.java:612)
>     at org.apache.geode.distributed.internal.InternalLocator.startLocator(InternalLocator.java:373)
>     at org.apache.geode.distributed.LocatorLauncher.start(LocatorLauncher.java:681)
>     at org.apache.geode.distributed.LocatorLauncher.run(LocatorLauncher.java:588)
>     at org.apache.geode.distributed.LocatorLauncher.main(LocatorLauncher.java:210)
> {noformat}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)