Error trying to synchronize cluster JNDI tree when starting managed server WebLogic 12.2.1.4

I'm updating an application from WebLogic 12.1.3 to 12.2.1.4. Our web application is on a cluster with 2 nodes. When we start the 2 managed servers, whichever one finishes last throws the following exception:

<Notice> <Cluster> <BEA-000138> <Listening for announcements from cluster WEBCluster on 239.255.0.102:10034.>
<Notice> <Cluster> <BEA-000133> <Waiting to synchronize with other running members of WEBCluster.>
<Notice> <Log Management> <BEA-170027> <The server has successfully established a connection with the Domain level Diagnostic Service.>
<Notice> <Cluster> <BEA-000142> <Trying to download cluster JNDI tree from server web-dev01.>
<Error> <Cluster> <BEA-000140> <Failed to deserialize statedump from server web-dev01 with
weblogic.application.ClassLoaderNotFoundException: com.app.ejb.SummaryRemote is not found due to missing GenericClassLoader.annotation:app-ejb@.
weblogic.application.ClassLoaderNotFoundException: com.app.ejb.SummaryRemote is not found due to missing GenericClassLoader.annotation:app-ejb@
    at weblogic.application.internal.AppClassLoaderManagerImpl.loadApplicationClass(AppClassLoaderManagerImpl.java:229)
    at weblogic.common.internal.ProxyClassResolver.resolveProxyClass(ProxyClassResolver.java:77)
    at weblogic.common.internal.WLObjectInputStream.resolveProxyClass(WLObjectInputStream.java:88)
    at java.io.ObjectInputStream.readProxyDesc(ObjectInputStream.java:1854)
    at java.io.ObjectInputStream.readClassDesc(ObjectInputStream.java:1802)
    Truncated. see log file for complete stacktrace

The server that finishes first shows the following with no issues:

<Notice> <Log Management> <BEA-170027> <The server has successfully established a connection with the Domain level Diagnostic Service.>
<Notice> <Cluster> <BEA-000138> <Listening for announcements from cluster WEBCluster on 123.123.0.102:10034.>
<Notice> <Cluster> <BEA-000133> <Waiting to synchronize with other running members of WEBCluster.>
<Notice> <WebLogicServer> <BEA-000365> <Server state changed to ADMIN.>
<Notice> <WebLogicServer> <BEA-000365> <Server state changed to RESUMING.>
<Notice> <Cluster> <BEA-000162> <Starting "async" replication service with remote cluster address "null">
<Notice> <WebLogicServer> <BEA-000330> <Started the WebLogic Server Managed Server "web-dev01" for domain "domain" running in production mode.>

Summary.java

package com.app.ejb;

@Stateless(mappedName = "Summary")
public class Summary implements SummaryRemote {
	@Inject
	private SummaryLocal summaryLocalBean;

SummaryRemote.java

package com.app.ejb;

@Remote
public interface SummaryRemote {

Lookup

private void getSummaryFacade() throws NamingException {
    Context context = new InitialContext();
    summaryRemote = (SummaryRemote) context
            .lookup("Summary#com.app.ejb.SummaryRemote");
}

Why is it throwing the ClassLoaderNotFoundException on the node of the cluster that starts up after the first one? I don't have weblogic-ejb-jar.xml or ejb-jar.xml set up for my project by the way.

Looking at the JNDI tree in the weblogic console, I see the same classloader not found exception when I have only one server up. I think when I bring server 2 up, it's trying to replicate/download the JNDI tree from server 1 and is unable to because of the error.
See link for image of the JNDI tree exception

A couple of work arounds that worked for me, doing either one got rid of the exception:

1. Change the cluster messaging mode from Multicast to Unicast in the WebLogic admin console.

Navigate to Environment -> Clusters -> <ejbnodename> -> Configuration tab -> Messaging tab -> Messaging Mode: Multicast, you need to press Lock & Edit and make changes from Multicast to Unicast and save and then release the Lock.

2. Updating weblogic-ejb-jar.xml to make the bean not clusterable.

    <wls:weblogic-enterprise-bean>
        <wls:ejb-name>Summary</wls:ejb-name>
        <wls:stateless-session-descriptor>
            <wls:stateless-clustering>
                <wls:home-is-clusterable>false</wls:home-is-clusterable>
                <wls:stateless-bean-is-clusterable>false</wls:stateless-bean-is-clusterable>
        </wls:stateless-clustering>
    </wls:stateless-session-descriptor>
   

   </wls:weblogic-enterprise-bean>

this might bug 30295025 in weblogic. Had a similar issue and after patch 30295025 was applied the issue went away.

Doc ID 2689913.1