Indexer Problems (AREV Specific)
At 13 AUG 1998 04:20:55PM Ed Bell wrote:
We are running AREV 3.12 on a NT network using the Revelation NT Service. I have set up a dedicated indexer but sometimes the indexer just stops indexing for no apparent reason. If I close the application and restart it the indexer starts indexing again and of course catches everything back up but I don't ever know that it has stopped until someone calls and complains about not finding something that they know has been enterd when they run a search.
I also have another client who is experiencing the same problem.
Has anyone else had this problem and if so how can it be fixed?
At 13 AUG 1998 08:56PM Chris Vaughan wrote:
I have come across this before, and in spite of some reasonable effort was never able to resolved it.
I seem to recall that the Indexing Server was looping in F.INDEXER (or DISTRIBUTOR or some such program) waiting for a record lock (cannot recall the file or record). I had installed a bespoke MFS and a shell around F.INDEXER to help see what was going on.
I thought that some other workstation had jumped in to do some index updates, perhaps via an explicit INDEX.FLUSH. I could not find anything suspicious in the application code, so I assumed that the contention was from within the bowels of AREV. Why the lock was never being released is the ongoing mystery.
The problem seemed to go away when the client changed their archiving policy (ie. they moved a lot of historical data into a secondary non-indexed table). I think there might be some fundamental problems with Btree indexing on large files with long multipart keys. The system gets into some kind of node management knot.
I don't work with Arev that much these days, but I would like to follow this up - so if you have any ideas about taking it further, I would be happy to do some of the work.
At 17 AUG 1998 03:48PM Ralph Johler wrote:
We are having similar problems on Novell v4.11 with the LH NLM. Our Indexer stops during a R/Basic batch job that uses Reduce and Select…By to extract data from a large (million records) file with a 4 part key of about 25 bytes. It appears that this problem is tied in with the Reduce/Select…By executing and Index.Flush (even though in this case it is not needed at all).
After running about 1/3 to completion our R/Basic batch job will also stop processing along with Indexer. Then sometimes the Indexer just loses the LH TSR to NLM connection and stops processing.
This is during the day, so it is not related to the BIOS on the PC.
Any other ideas?