topic Re: Lucene Index Backup in Alfresco Archive

Lucene Index Backup

the_medina — Tue, 03 Aug 2010 12:43:01 GMT

Hello,Hopefully someone here may be able to assist me, as I am almost entirely out of ideas. I am attempting to develop a backup / restore plan for the company I work for. I have found great aid in this link http://wiki.alfresco.com/wiki/Backup_and_Restore . Everything appears to make sense thus f

Re: Lucene Index Backup

zaizi — Tue, 03 Aug 2010 16:23:53 GMT

I have a strong suspicion that triggering lucene index backups via JMX is not available in Alfresco 2.1.

You can confirm this by viewing the JMX options in your location installation. See http://wiki.alfresco.com/wiki/JMX.

Ainga

Re: Lucene Index Backup

m2acis — Thu, 19 Aug 2010 10:49:22 GMT

I'm also struggling with implementing Alfresco backup and (apparently) I'm having trouble with Lucene indexes too.

In my test environment I do CDP replication of whole "C:\Alfresco" from primary to remote server. When I restore data on remote server DB (and filesystem) is consistent. Documentation states that lucene-indexes are backed up daily at 3am (http://wiki.alfresco.com/wiki/Backup_and_Restore). So if I move "backup-lucene-indexes" to "lucene-indexes", as advised in docs, then Alfresco should update indexes on startup (index.recovery.mode set to "AUTO").

But it appears that indexes are not updated or not updated correctly. I have uploaded some word and openoffice documents yesterday. Alfresco was running all night and "lucene-indexes" were backed up at 3:00am. Today I uploaded "notes.odt" at 11:02 and it contains word "netbackup". If I enter "netbackup" in search, alfresco on remote server doesn't show "notes.odt" while on primary server it finds "notes.odt". Of course I can see all files in my "Dokumenti" space.

If I skip the step of moving "backup-lucene-indexes" to "lucene-indexes", then search works. But this is equal to backuping of "lucene-indexes" with Alfresco running and docs say that "doing so is almost certain to cause Lucene index corruption." Setting index.recovery.mode set to "FULL" doesn't help neither.

What could be wrong here? What about that "almost certain to cause Lucene index corruption" thing? Does it also apply if I do filesystem snapshots?

Running Community v3.3.0 (2765) on win 2003 server r2

Text from logs, when "index.recovery.mode=AUTO" and "backup-lucene-indexes" renamed to "lucene-indexes"

12:24:00,713 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent] Index recovery started.
12:24:02,276 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent]    Processing transactions around Thu Aug 19 11:32:46 EEST 2010.
12:24:06,463 INFO  [org.alfresco.repo.node.index.FullIndexRecoveryComponent] Index recovery completed.‍‍‍

Re: Lucene Index Backup

m2acis — Fri, 20 Aug 2010 07:57:25 GMT

Ok! It looks like I'm missing some key understanding about Lucene indexes in Alfresco and how they work. I just did a test on primary server too - stopped Alfresco, renamed "lucene-indexes" to "original-lucene-indexes" and started Alfresco (index.recovery.mode=AUTO). Alfresco perforemed full index recovery, but search does not work… Does testing the indexes using "search" is the right way?