https://jira.opendaylight.org This file is an XML representation of an issue en-us 8.20.10 820010 22-06-2022 https://jira.opendaylight.org/browse/CONTROLLER-1299 controller <p>CDS get's restarted when it's configuration is changed using NetConf or when a NetConf mount is added. In this scenario CDS does not restart properly because the akka actor system that is used by CDS gets terminated on stopping any one of the datastores and does not get recreated when the datastore is recreated.</p> <p>Operating System: All<br/> Platform: All</p> CONTROLLER-1299

Clustering : CDS fails to restart when configuration changes

Bug Resolved Done Moiz Raja Moiz Raja Fri, 8 May 2015 00:35:20 +0000 Tue, 26 May 2015 16:19:29 +0000 Tue, 26 May 2015 16:19:29 +0000 Post-Helium mdsal 0 1 <p>I created <a href="https://bugs.opendaylight.org/show_bug.cgi?id=2976" class="external-link" target="_blank" rel="nofollow noopener">https://bugs.opendaylight.org/show_bug.cgi?id=2976</a> for this same issue. This issue also occurs if you restart the controller. There's actually 2 issues here. The controller restart case is indirectly caused by the feature config pusher and persistent config pusher both trying to start the modules. Having both pushers doing the same thing on startup could lead to system instability. The underlying cause is that CDS uses a static ActorSystem and thus it tries to reuse it when restarted in the same JVM instance.</p> <p>I think we need to just null out static field when closed.</p> <p><a href="https://git.opendaylight.org/gerrit/19957" class="external-link" target="_blank" rel="nofollow noopener">https://git.opendaylight.org/gerrit/19957</a></p> <p>I've tried putting in a more comprehensive fix here. The reproduction steps I had used are,</p> <p>1. Bring up controller<br/> 2. Install the netconf feature<br/> 3. Add a netconf mount<br/> 4. Restart the controller<br/> 5. Check the logs and the mbeans. </p> <p>In the logs you will see an exception that you are trying to create an actor on an actor system that is terminating/terminated. In the MBeans you will see that one of the Distributed*DataStores will not be created. The system may continue to work because the IMDS may be used instead of CDS but in that case persistence will be broken.</p> <p>Yes - those are the issues I saw.</p> <p>(In reply to Moiz Raja from comment #3)<br/> &gt; I've tried putting in a more comprehensive fix here. The reproduction steps<br/> &gt; I had used are,<br/> &gt; <br/> &gt; 1. Bring up controller<br/> &gt; 2. Install the netconf feature<br/> &gt; 3. Add a netconf mount<br/> &gt; 4. Restart the controller<br/> &gt; 5. Check the logs and the mbeans. <br/> &gt; <br/> &gt; In the logs you will see an exception that you are trying to create an actor<br/> &gt; on an actor system that is terminating/terminated. In the MBeans you will<br/> &gt; see that one of the Distributed*DataStores will not be created. The system<br/> &gt; may continue to work because the IMDS may be used instead of CDS but in<br/> &gt; that case persistence will be broken.</p> Duplicate CONTROLLER-1298 Development External issue ID 3156 External issue URL Issue Type ODL SR Target Milestone Priority Rank 0|i02ppr: