Hi
Have a bunch of processes connected to Postgres (10.0) on windows server
2016. The software was running without problems for at least a year.
Each process is doing different actions.
Every about 5-10 days a postgres.exe running for one of the application
processes crashes, tears down everything. Database recorves and then it
works again as normal.
The wierd thing is it's not always the same application process which
starts the whole sequence.
One success situation is as follows
2021-07-12 07:33:52.701 CEST [5744] LOG: duration: 0.020 ms statement:
COMMIT
2021-07-12 07:33:52.707 CEST [5744] LOG: duration: 0.023 ms statement:
DISCARD ALL
2021-07-12 07:33:52.707 CEST [5744] LOG: duration: 0.005 ms statement:
BEGIN
2021-07-12 07:33:52.707 CEST [5744] LOG: duration: 0.011 ms statement:
SET TRANSACTION ISOLATION LEVEL READ COMMITTED
.....
The same potgres.exe crashes afterwards
2021-07-12 07:33:55.321 CEST [5744] LOG: duration: 0.032 ms statement:
COMMIT
2021-07-12 07:34:32.660 CEST [3972] LOG: server process (PID 5744) was
terminated by exception 0xFFFFFFFF
The COMMIT was the last successfull action in the database. I verified
the existence of the data in the database.
The new job at 07:34 caused the crash immediately.
Since I dont see DISCARD ALL/BEGIN/SET TRANSACTION my interpretation is
this:
The possible (if any) corrupted shared memory is caused by an other
postgres.exe process.
Therfore checked all actions before these crashes. But also her it's
always a different one.
BTW: There is no virus scanner on the postgres folder and it's a
productive system, I cant just update.
So the problem leaves me puzzled.
Would be great if somebody has some ideas how to continue investigation.
Thanks
Beat
