Disable L2 cache

[jsb]

We have migrated our Alfresco instance (3.4.11 Enterprise) into AWS. The EhCache automatic discovery is failing since AWS does not support multicast. To fix this I changed the ehcache-context.xml file to use manual discovery (default was automatic discovery). This works just as well as the automatic discovery except for when one of the Alfresco instances in the cluster goes down (server issue or tomcat restart). When this happens Share and Explorer for all the other servers in our cluster become unusable.

Is there a way to completely disable the L2 cache? Just pull the information from the database? We can test the performance hit in this situation if it is possible.

- or -

Do you have any ideas about how to configure the EhCache to to use manual discovery and not degrade the Alfresco instance if another instance in the cluster is down?

And of course if you have any other helpful ideas please let me know.

[kaynezhang]

Try to add this JVM parameter to your app. server:

-Dnet.sf.ehcache.disabled=true

But I'm not sure whether it will cause any errors,you should give it a test.

[jsb]

This didn't seem to help, but thank you for pointing it out. I am still investigating why it isn't working, because it seems like exactly what I want.

[kaynezhang · Accepted solution]

I don't have source code for version 3.4 ,maybe you can try to set the following properties to true in the alfresco-global.properties :

system.cache.disableMutableSharedCaches=true

system.cache.disableImmutableSharedCaches=true

It works for 4.0,maybe it will also work for 3.4 you can give it a try.

[jsb]

Well that did seem to disable caching, but it made the servers just as slow as when there is a server down in the list.

But I guess I'll mark this as the correct answer since it did answer the question...

[afaust]

Are you an active Alfresco Enterprise customer? If so, you should also consider contacting Support with that question. Of course they will very likely tell you to upgrade to a newer Alfresco version as 3.4.x is quite old and no longer fully supported. There also has been quite a bit of change in the clustering area with Hazelcast now being used for clustering and discovery. Though that too has some issues, overall the cluster stability was improved greatly with the switch to Hazelcast-based clustering.

[jsb]

Thank you for the reply. Support was not able to help much with the 3.4 version. We know that upgrading would fix a lot of small issues and probably help with performance. Hopefully that will come soon.

[jsb]

We ended up just moving forward with this 'as is'. We'll just fix the config file whenever the issue occurs, which is rarely.

Also, I made a mistake in my original post, this only happens when a server is down, not when tomcat is down.

Thank you for your help everyone.