Dedicated Indexer (OpenInsight 32-bit Specific)
At 31 JAN 2006 06:24:04AM Simon G Wilmot wrote:
Can anyone tell me what the indexer is doing when,
a) The indexes are all up to date.
b) The indexer is the only PC running on the network …
... and is the only thing running on that PC.This is a Novell network, NLM running, OI v4.02 - watching the system screens the indexer is writing to each Bang file in the cycle. This also means on its cycle it also manages to get the CPU utilisation up to between 30% & 40% !!
This would appear to be doing excessive work …
Regards,
Simon
At 31 JAN 2006 01:02PM Karen Oland wrote:
We run a dedicated indexer in the AREV environment - it can push the server up to 50% utilization if you let it. Instead, we wrote a routine to do indexing – it builds a list of indexed files, then updates each one, one at a time, then waits (5 seconds or more) before doing the next one (it also attaches different locations before repeating, which you probably don't need).
Once indexes were caught up, it spends most of it's time waiting and no longer bangs the server to death reading the 0 records in !files.
At 31 JAN 2006 01:05PM Karen Oland wrote:
We run a dedicated indexer in the AREV environment - it can push the server up to 50% utilization if you let it. Instead, we wrote a routine to do indexing – it builds a list of indexed files, then updates each one, one at a time, then waits (5 seconds or more) before doing the next one (it also attaches different locations before repeating, which you probably don't need).
Once indexes were caught up, it spends most of it's time waiting and no longer bangs the server to death reading the 0 records in !files.
At 31 JAN 2006 03:58PM support@sprezzatura.com wrote:
OI 4.02 had some issues with the timers. You might want to try 4.03.
support@sprezzatura.com
The Sprezzatura Group Web Site
World Leaders in all things RevSoft
At 01 FEB 2006 10:50AM Simon G Wilmot wrote:
We will try that, but do you know what it is actually writing. I did a binary compare of a changed file after an indexers sweep and the two files were identical …
Simon
At 01 FEB 2006 12:18PM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Most likely record 0 with a field mark.
World leaders in all things RevSoft
At 02 FEB 2006 05:00AM Simon G Wilmot wrote:
Having checked, that is the case. But why ?? Surely this is an unnecessary overhead !!
Simon
At 02 FEB 2006 07:49AM [url=http://www.sprezzatura.com]The Sprezzatura Group[/url] wrote:
Oversight.
World leaders in all things RevSoft
At 02 FEB 2006 10:01AM Simon G Wilmot wrote:
A feature then