On 27/08/2009 18:11, Martin Pihlak wrote:
>>> There are actually no "real" data changes being made on your master
>>> for some reason. So every time archive_timeout is reached a log full
>>> of no changes is shipped to your slave and applied - and no checkpoint
>>> times are changed for reasons I mentioned above.
>>>
>>>
>>>
>> thanks, but we have not set archive_timeout, and we have a lot of real
>> data changes.
>>
>> That's why i don't understand why checkpoint never happen on the slave.
>>
>>
> What about the other values in pg_controldata output, do they change at
> all? If they remain constant, maybe the logs are just not applied. Also,
> are there any "restored log file xxx from archive" entries in postgres
> log?
>
>
>
No, i don't see any change in pg_controldata, but :
there is messages in logfile :
2009-08-28 10:02:51,129 26717 INFO 00000001000003F700000088: Found
2009-08-28 10:02:51,169 26717 INFO {count: 1}
2009-08-28 10:02:51 CEST [18439]: [1862-1] user=,db= LOG: restored log
file "00000001000003F700000088" from archive
postgres@bdd2:/data/postgresql/8.3/main$ grep "restored log file"
/var/log/postgresql/postgresql-2009-08-28.log | wc -l
164
And I can see that the cluster is updated inside the directory px_xlog
and base, at least.
So we can think it's working, but, when i restart the slave cluster, it
process again all the WAL since the last checkpoint ( for nothing, I
think ), so, when we'll need this slave, it will take a lot of time,
more and more, and this point worried me a bit.
--
Sébastien Lardière