[Thread-SIG] RE: tstate invalid crash with threads

Ray Loyzaga ray at commsecure.com.au
Wed Jun 9 18:43:56 EDT 1999


Scott Cothrell wrote:
> 
> I can confirm that we were seeing similar behavior under a highly threaded,
> highly stressed situation.  We backed off on the number of threads and
> slowed things down as a workaround also.  We couldn't trace it down to
> Python, as we have a Java, C, Python mix that has too many suspects.
> 

The locking code added to the create/destroy side of the threads
has yielded good results. On my stripped down server, which was crashing
after between 40,000 and 1.4M transactions managed 14.75M last night.
I will kill this and attempt to run my real task for a day or so,
but it seems to confirm a small race exists ....




More information about the Python-list mailing list