We seem to have a problem with search on our systems. Essentially a shutdown of the server hangs saying “Waiting for search server to shutdown”, requiring a manual stop of the execd.exe process.
Also we see searches for related content simply hanging for potentially hours. The same with an incremental publish, which I assume must use search more than a full publish because a full content list returns quickly, an incremental simply hangs the system.
Can you advise a way to completely clear the search indexes (deleting files from the sys_search folder perhaps), followed by a re-index (search index type monitor command)? I think one of the issues is that search was originally installed on the C: drive of the Land Sec systems we copied over and despite reconfiguring this in the System administrator to point to our F: and changing any references in config files within the sys_search folder itself, it simply doesn’t seem to be happy.
I’ve raised a support call (UK) about this but any other info would be appreciated…
Here are some instructions that I got via support for cleaning out the indexes of Convera and starting afresh.
[ul]
[li]Stop your Rhythmyx Server if it’s running
[/li][li]Go To <Rhythmyx_root>\sys_search\rware\rx\indexes and delete everything in that folder
[/li][li]Open your Rhythmyx database and completely truncate the PSX_SEARCHINDEXQUEUE table
[/li][li]Start your Rhythmyx Server
[/li][li]Launch the Rhythmyx Server Administrator
[/li][li]Go to the Monitor tab and type in the following command: “search index type”
[/li][/ul]
NOTE: This will completely reindex all of your content and will probably take several hours to complete depending on your environment so please plan accordingly. You can watch the progress by typing the following command: “show status search” into the console. This will tell you how many items are queued for indexing.
I keep seeing the following in my console.log though, although will start the reindex over the weekend…
2008-02-08 19:07:25,062 INFO [PSServerLogHandler] appid: 0 type: 1 time: Fri Feb 08 19:07:25 GMT 2008 msg: An internal error (end-condition) was encountered. An unexpected exception occurred:
An unexpected exception occurred. The reason was: The search engine ‘Convera’ failed to connect to the server. The specific problem is: server failed… Please consult the log for further information…
The call-stack was:
com.percussion.search.convera.PSConveraSearchException: The search engine ‘Convera’ failed to connect to the server. The specific problem is: server failed.
at com.percussion.search.convera.PSIndexContext.commitIndex(Native Method)
at com.percussion.search.convera.PSIndexContext.super(Unknown Source)
at com.percussion.search.convera.PSSearchIndexerImpl.update(Unknown Source)
at com.percussion.search.i.super(Unknown Source)
at com.percussion.search.i.super(Unknown Source)
at com.percussion.search.i.super(Unknown Source)
at com.percussion.search.i.super(Unknown Source)
at com.percussion.search.i$1.run(Unknown Source)
2008-02-08 19:07:25,078 ERROR [FTSIndexQueue] An unexpected exception occurred. The reason was: The search engine ‘Convera’ failed to connect to the server. The specific problem is: server failed… Please consult the log for further information.
2008-02-08 19:07:25,468 INFO [i] processing event: [40068]:
<?xml version=“1.0” encoding=“UTF-8”?>
<PSXEditorChangeEvent action=“reindex” childId="-1" childRowId="-1" contentId=“1143” contentTypeId=“326” priority=“100” revisionId=“1”/>
2008-02-08 19:07:25,625 INFO [PSSearchEngineImpl] ‘status’ completed with return code = 0
2008-02-08 19:07:25,625 INFO [PSSearchEngineImpl] ‘status’ console output:
Does rebooting the Rx server have any impact on the behavior? Was this working and just stopped working one day, or did it get progressively worse?
Generally, the ‘server failed’ type of error means that one or more of the Convera processes failed to start or hung. Check the logs in [rxinstall]/sys_search/rware/rx/logs to see if any message was logged there. If not, then you will need to attempt to start the programs from the command line, which requires more work.
We have a situation where items sre stuck in search index queue but I can’t find them if I search using content Id field in Advance Search. Items are in contentstatus table, in Public state.
If you have Convera FTS turned on and the item hasn’t been indexed, you will not find it using any field in the search. I would recommend clearing your indexes and then re-indexing your environment. Please open a ticket with TS to get the doc with the full steps.
Do you have more than one Rhythmyx installation running on the same network?
Most of my Convera application failures we due to this and assigning a different port for each to listen on in rxconfig\server\config.xml cleared up a lot of issues.
Dan is right - this sounds like multiple Rx instances running Convera with ports colliding and self terminating connection.
One more thing to check - see the server log if you see any loops trying to find objects and timing out - this may be caused by recent cleanup of the system(if you have done so) and orphaned items as a result that are indexed but can’t be found.
i.e. - deleted Content Types, without purging the content items etc.
