Hi guys!
I use pgsql for some time already and am happy with it. Heh, sure this
post has it's big BUT :-)
Starting few months ago, one of our projects encoutered loss of one DB
table (in that time, it was version 8.0 or so...) I did some research
and found out, the vacuuming was set wrong and transaction wraparound
failure occured... OK, I upgraded to 8.2.1 (latest in that time) and set
the auto vacuum... (till now I think it was set correctly and
working)... but after less than one month, the same happened... but this
time, the age of all databases looked OK...
So I did upgrade to 8.2.3 (the project is running this version till now)
but the same happened today...
This time I didn't rely on the autovacuuming and did code cron scheduled
job to do vacuum full on all tables during nightly maintenance. I'm now
sure, the wraparound didn't occure (am I?)... in fact, I'm lost...
One more thing:
The project runs proprietal CMS system and there are more instances of
it with the same database layout in different databases. Every time the
"lost" table is the same one - the bussiest one (mostly read)... and
everytime the "lost" table is gone from more projects at once... today
it was 2 projects affected out of 7 in the same pgsql server... it looks
also like the problem happens everytime short after midnight - scarry -
ha ha :-)
I have the backup of DB files as they was this morning, but it's alter
the vacuum script...
My system is:
OS: Debian 3.0
Kernel: 2.4.26
PgSQL: 8.2.3
Any advice and/or hint to the right direction is greatly welcome!
Regards,
--
Kamil
---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend
