[ic] interchange crashing?

[Mike Heins]

Quoting John Beima (jbeima@palb.com):
> We have had a fair number of unique crashes as well, as of late...
> 
> They all occur with the new "rpc" mode... It is kinda very funny... We have 200
> sites on one machine, only one of which is based on the foundation template...
> It is the ONLY catalog that crashes... All of our custom made ones run fine...
> 
> We have another server with 18 custom made catalogs and 1 continues to freeze...
> 
> The interesting thing is is you alter the "rpc" settings to allow each server to
> process no more than 40 requests instead of 100 the crashes seem to totally
> dissappear... They have gone from 6 or 7 locks a day to 4 days with no
> crashes... My buess is there is a memory leak in "rpc" mode... The change we
> made was to start up 10 servers instead of 5 because of the drops in requests
> allowed to each server...
> 
> Hope this helps someone out...

I think what is happening is that DBI is caching the MySQL handles
and at some point they stop working.

If anyone has seen this problem with DBM-only or Postgres sites (neither
of which cache connections) then I would appreciate hearing about it.
I would really like to nail this problem.

-- 
Red Hat, Inc., 3005 Nichols Rd., Hamilton, OH  45013
phone +1.513.523.7621      <mheins@redhat.com>

I am a great believer in luck, and I find that the harder I work
the more luck I have. -- Thomas Jefferson