topic 2 servers in a cluster, file replication not 100% in Alfresco Archive

2 servers in a cluster, file replication not 100%

usuf — Wed, 04 Nov 2009 16:44:29 GMT

Hello,I am in the file replication configuration stage of setting up 2 servers in a cluster. One is located at head office, the other at a branch in another city.I'm using Alf community 3.2 on fedora 10 machines with multicast running and index rebuilding set to AUTO every 5 seconds. There is a VPN

Re: 2 servers in a cluster, file replication not 100%

usuf — Wed, 04 Nov 2009 18:45:07 GMT

Update: If it helps, when I upload a file using webDav, it appears in full, in both locations. I added a description to the file at Head office, I then refreshed Guest Home at the branch office, and the file reverted back to having only a file name and a creation date. I could not view it in a browser anymore.

Re: 2 servers in a cluster, file replication not 100%

chapeaurouge — Wed, 11 Nov 2009 11:13:42 GMT

Are you on 3.2r?

I had this working fine on release 3.2, and now I get bunch of

10:34:20,019  DEBUG [node.index.IndexTransactionTracker] Reindexing batch of 1 transactions from Wed Nov 11 10:30:53 CET 2009 (txnId=4323)
10:34:20,035  DEBUG [node.index.IndexTransactionTracker] Reindex skipping transaction: Transaction[id=4323, txnTimeMs=2009-11-11T10:34:15.876+01:00, changeTxnId=e86a130e-19ad-431b-9bc9-83c82fc89d9d]
10:34:20,035  DEBUG [node.index.IndexTransactionTracker] Voids detected: 1 in range [4321, 4323]
10:34:20,037  DEBUG [node.index.IndexTransactionTracker] Void count 8 -> 7
10:34:25,018  INFO  [node.index.IndexTransactionTracker] reindexImpl started: org.alfresco.repo.node.index.IndexTransactionTracker@12ee90d9
‍‍‍‍‍‍

So my indexes are not replicated between my 2 nodes.. I see stuff going on between the 2, but nothing is up-to-date on the 2nd node.

Re: 2 servers in a cluster, file replication not 100%

tommorris — Wed, 11 Nov 2009 12:16:20 GMT

Could it be that the two servers' clocks need to be synchronised?

Re: 2 servers in a cluster, file replication not 100%

chapeaurouge — Wed, 11 Nov 2009 12:48:37 GMT

No, I checked that. Both via NTP, sync'ed up to the second at least.