[Building Sakai] still struggling with indexer errors

[Matthew Jones]

Are you using the SearchService mode for search? [1] At Michigan we've
always been somewhat wary of search and have it stealthed overall and have
one single machine out of the cluster dedicated as the search server. All of
the other machines use this one. It seems like turning it on for everyone
without an easy way to limit what is being indexed

We also have it only indexing the tools with search placed
"onlyIndexSearchToolSites at org.sakaiproject.search
.api.SearchIndexBuilder=true"

As an aside it seems like it would be nice to further limit or exclude
specific the documents searched (like excluding pdf, or reducing the max
file size of content hosting documents searched - SRCH-24). We may also want
to just exclude certain tools from being indexed at all.

This possibly could be achieved through some server properties pretty easily
in the same section that disables search for the adaptor.

(Note: These don't actually do anything)

search.index.content = true

search.index.mailarchive = true

Perhaps this *might* still be able to be enabled/overridden on a per site
basis at runtime with a site property

All of this would be a decent amount of work though. And there was some
rumblings at one time about replacing search with something else (don't
know/remember the status of that)

Good luck! I've seen some of these errors on a local QA server and even
cleaning them up didn't resolve it. It seems like "searchServer" has been
much more reliable though.

[1]
http://confluence.sakaiproject.org/display/SEARCH/Home#Home-SearchServer%2FSearchClient
 On Thu, Oct 28, 2010 at 5:01 PM, Joshua Swink <joshua.swink at gmail.com>wrote:

> We're still struggling with the indexer errors that say "no segments*
> file found", as in the ticket
> http://jira.sakaiproject.org/browse/SRCH-15. I'm just throwing this
> out there in the hope that someone familiar with the code or the
> configuration of the search tool may have some idea of how to correct
> this.
>
> Our last effort to fix this was as follows:
>
> * Verify that sharedJournalBase at org.sakaiproject.search.api.JournalSettings
> pointed to a shared location that all tomcat instances have read/write
> access to.
> * Verify that localIndexBase at org.sakaiproject.search.api.JournalSettings
> points to a local location for each tomcat instance, which is not
> shared.
> * Shut down all Tomcat instances, then:
> ** Erase the contents of the sharedJournalBase directory.
> ** Erase the contents of all search tables in the database:
> search_journal, search_node_status, search_segments,
> search_transaction, searchbuilderitem, and searchwriterlock.
>
> After starting Sakai up again, the errors continue, on all instances.
> Also note that search does not work - any attempt to perform a search
> generates a similar error. The search indexer is constantly hitting
> the database pretty hard, making up the bulk of the database load for
> Sakai.
>
> Josh
> _______________________________________________
> sakai-dev mailing list
> sakai-dev at collab.sakaiproject.org
> http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
>
> TO UNSUBSCRIBE: send email to
> sakai-dev-unsubscribe at collab.sakaiproject.org with a subject of
> "unsubscribe"
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/sakai-dev/attachments/20101028/7f926517/attachment.html