Cluter Configuration - How do you make it work?

As we transition to the new Hyland Community, some post links may be temporarily disrupted as we finalize redirects. We apologize for the inconvenience and are working to resolve this as quickly as possible. Details.

machtech

Hi guys,

I am running a 2-headed cluster on Alfresco Community 3.3g on Ubuntu 10.04 using a shared content store (via NFS) and a shared database. I have followed all the instructions at http://wiki.alfresco.com/wiki/Cluster_Configuration_V2.1.3_and_Later.

The problems I am having are numerous but, while simple content changes appear to replicate:
  1. Searches on both head nodes return different results (indexes not replicating?)
  2. Large deletions from the repository don't replicate (one node thinks the data is there, the other does not)
  3. Permissions do not replicate from one node to another (and an ACE here, it doesn't show up there)
  4. Customise the user dashboard in Share and it does not replicate to the other node

What exactly is it that I am doing wrong?

Here's the important properties from alfresco-global.properties from one of the nodes:

=======================================================================
# Cluster configuration
cluster.node=node1
alfresco.cluster.name=ecm

# Content and index locations
dir.root=/data/alfresco_data
dir.indexes=${dir.root}/${cluster.node}-indexes/lucene-indexes
dir.indexes.backup=${dir.root}/${cluster.node}-indexes/backup-lucene-indexes

# Index configuration
index.tracking.cronExpression=0/5 * * * * ?
index.recovery.mode=AUTO
=======================================================================
‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍‍

The /var/lib/tomcat6/shared/classes/alfresco/extension/ehcache-custom.xml is simply copied from the .sample file in the 3.3g distribution.

The time on the boxes is in sync, NTP is installed on both, they are spot on.

What is going on?

Cheers,
Johan

machtech

Hi again,

At least for some of the issues up there I discovered the culpruit - I was using MySQL replication with two master write notes replicating to each other, and when I did a big delete (1000s of items) on one of the nodes, something bad happened causing the replication to pause between the database servers. This caused some of the behaviour I mentioned, but doesn't account for all of it - the Share dashboards not updating on both nodes still occurs (I have lodged a bug for this a while ago but no responses yet).

Anyway, leaving the post up there for posterity and Google.

Cheers,
Johan