topic Re: Question with: the SOLR indexes and repository database do not match in Alfresco Archive

Question with: the SOLR indexes and repository database do not match

carlos_cruz — Mon, 06 May 2013 22:12:34 GMT

Hi everyone, I am a new alfresco user, and my alfresco log is always writing:<break>May 6, 2013 12:00:00 AM org.alfresco.solr.tracker.CoreTracker trackRepositorySEVERE: First transaction was not found with the correct timestamp.May 6, 2013 12:00:00 AM org.alfresco.solr.tracker.CoreTracker trac

Re: Question with: the SOLR indexes and repository database do not match

mitpatoliya — Tue, 07 May 2013 09:12:36 GMT

First you can try with full indexing.
Then if it still does not work delete the existing solr indexes as per your steps and then enable full indexing to rebuild all the indexes. do not forget to restart the server.
For full re-indexing you need to change following property in your alfresco-global.properties file.
index.recovery.mode from value "AUTO" to "FULL"

Re: Question with: the SOLR indexes and repository database do not match

carlos_cruz — Tue, 07 May 2013 15:44:20 GMT

Thanks a lot mitpatoliya !!, but I don't know if this procedure is secure. If I do that, my data and documents will be safe?
I need to be sure because my Alfresco has real data.

thank you all for your time

Re: Question with: the SOLR indexes and repository database do not match

bopolissimus — Tue, 07 May 2013 23:36:20 GMT

Hi Carlos,

Generally, rebuilding the indexes is safe. *While* the index is rebuilding some functionality that depends on search won't work (or may work inconsistently as index updates for some documents are completed while others aren't), the content and metadata themselves aren't primarily stored in the indexes so it'll be safe.

On the other hand, if your alfresco setup (including data) is important, then you should really have a scheduled and TESTED backup and restore procedure. If you don't have that then hardware failure will have you lose your content or metadata and you'll have a hell of a time trying to recover your data.

Also, a backup and restore procedure lets you test things (like full reindexing) on another machine. If your content is important, it's very important to test features/functionality on a test/staging machine. Only apply tested procedures on production once you're happy with results on staging. Sometimes (a recent alfresco upgrade comes to mind), even thorough rehearsal on staging will not work exactly the same way when you perform the same procedure on production and then you might need to fix things in realtime on the production server if you're already halfway through your upgrade. In those cases it's still a good thing to have a backup since, if your upgrade fails halfway and you can't make progress, you can still revert to your old production setup.

One thing to note here is that index.recovery.mode=FULL does not have any effect in solr. It's only effective if your index.subsystem.name=lucene. For solr, there's a different procedure for regenerating the indexes from scratch. e.g.,

http://docs.alfresco.com/4.0/index.jsp?topic=%2Fcom.alfresco.enterprise.doc%2Ftasks%2Fsolr-reindex.html

(find the appropriate documentation page for your alfresco version).

Again. Don't perform fixes on production without first testing on a separate machine. I often test twice or more times just to make sure I've got the procedure down.

Gerald