On 05/30/2014 01:59 AM, snowsnake wrote:
> postgresql 8.1.11
> we found the database can't be connected at 2014/5/9 23:00 and 2014/5/10
> 17:03 which you can see from below log. We checked the CUP and memory also,
> it is normal. And the server generated many postmaster process at the
> interrupted time. Untill we restart the database, it becomes normal. So why
> this could happens?
See comments inline.
>
> DB log : 5/9 23:00 ~ 00:00
> tail postgresql-Fri.log
> [PID=16483]: [user=ec_operator,db=eservice,ip=10.100.66.37][time=2014-05-09
> 23:00:03 CST]LOG: duration: 0.024 ms
> [PID=16485]: [user=ec_operator,db=eservice,ip=10.100.66.37][time=2014-05-09
> 23:00:03 CST]LOG: statement: <BIND>
>
> head postgresql-Sat.log
> [PID=5263]: [user=,db=,ip=][time=2014-05-10 00:15:58 CST]LOG: database
> system was interrupted at 2014-05-09 23:01:04 CST
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
It would depend on what did the above.
More information would be helpful:
1) OS and version
2) Where did you get the information that many postmaster processes
where generated?
3) Any idea what those processes where doing?
4) Any scripts (cron or otherwise) running at these times?
>
> 5/10 19:00~20:00
> [PID=13060]: [user=ec_operator,db=eservice,ip=10.100.66.37][time=2014-05-10
> 19:03:21 CST]LOG: statement: EXECUTE <unnamed> [PREPARE: select
> currval('public.ec_messagebackup_id_seq')]
> [PID=13060]: [user=ec_operator,db=eservice,ip=10.100.66.37][time=2014-05-10
> 19:03:21 CST]LOG: duration: 0.032 ms
> [PID=19628]: [user=,db=,ip=][time=2014-05-10 20:52:24 CST]LOG: database
> system was interrupted at 2014-05-10 19:05:21 CST
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Just so I am clear this is the 17:03 event you refer to above?
--
Adrian Klaver
adrian.klaver@aklaver.com