Likely wrong interruption time reported in log files (8.0.0 and 8.0.3).

[Knut P Lehre]

For some reason the second part of my post ("1) ...  2) Likely wrong interruption
time reported in log files (8.0.0 and 8.0.3).") was split up into several
parts, and some of it disappeared. I am therefore resending it as a separate
message. Could be that the bugs list would be more appropriate. Anyway,
the following happened on Windows Server 2003.

The pgsql server was interrupted a few times, likely because of a hardware
problem. I wanted to check at what time the server had been interrupted,
and found the following info in the log files:

From 1st log file, restart after planned shutdown for backup at 05:00:00
(running version 8.0.0):

2005-04-21 05:55:15  LOG:  database system was shut down at 2005-04-21 05:00:00
W. Europe Daylight Time
2005-04-21 05:55:15  LOG:  database system is ready

From 2nd log file, automatic restart after likely hardware failure:

2005-04-21 07:07:32  LOG:  database system was interrupted at 2005-04-21
05:55:15 W. Europe Daylight Time
2005-04-21 07:07:32  LOG:  database system is ready

Note that interruption time is identical to the last startup time.
And similar some weeks later (after upgrade to 8.0.3):

From 1st file:

2005-05-31 05:57:52  LOG:  database system was shut down at 2005-05-31 05:00:00
W. Europe Daylight Time
2005-05-31 05:57:52  LOG:  database system is ready

From 2nd file:

2005-05-31 09:43:04  LOG  database system was interrupted at 2005-05-31
05:57:52 W. Europe Daylight Time
2005-05-31 09:43:04  LOG:  database system is ready

Again, the reported interruption time is identical to the last startup time.
The reported pg interruption times also did not match the timestamps produced
by other processes during these events. Could it be that the logger
uses a wrong time stamp for the interruption time report? Does it depend
on settings in postgresql.conf?



---------------------------(end of broadcast)---------------------------
TIP 5: don't forget to increase your free space map settings

[Tom Lane-2]

"Knut P Lehre" <k.p.lehre@...> writes:
> Again, the reported interruption time is identical to the last startup time.

I don't see anything wrong here --- the reported timestamp is just that
of the last checkpoint, and if the server wasn't up long enough to
create a new checkpoint then this is exactly what you'd expect.

                        regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend

[Knut P Lehre]

>>"Knut P Lehre" <k.p.lehre@...> writes:
>> Again, the reported interruption time is identical to the last startup
time.
>I don't see anything wrong here --- the reported timestamp is just that
>of the last checkpoint, and if the server wasn't up long enough to
>create a new checkpoint then this is exactly what you'd expect.
>
> regards, tom lane

I see. Thanks.



---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?

               http://www.postgresql.org/docs/faq