On Sep 7, 2004, at 20:03, Tom Lane wrote:
> I suspect that it's got nothing at all to do with the traffic between
> the server and the client, and that the SIGINT is coming from some
> outside agency. However, if you want to be sure about it, set the
> postmaster's log_min_messages to DEBUG2, and then the next time the
> problem happens, look in the postmaster log to see if you see anything
> about
> processing cancel request: sending SIGINT to process %d
Yes, you are correct. I added the debug flag and reproduced the
problem. This was written to the log:
<2004-09-08 08:55:43 CEST 413eacee.3402>DEBUG: processing cancel
request: sending SIGINT to process 13311
<2004-09-08 08:55:43 CEST >DEBUG: server process (PID 13314) exited
with exit code 0
<2004-09-08 08:55:43 CEST 413eace4.33ff>ERROR: canceling query due to
user request
<2004-09-08 08:55:43 CEST >DEBUG: forked new backend, pid=13315
socket=60
<2004-09-08 08:55:43 CEST 413eacef.3403>DEBUG: processing cancel
request: sending SIGINT to process 13311
Are the OS sending the SIGINT to PostgreSQL or are postmaster sending
it to one of its child?
Regards,
- Tore.