This is 8.2.6 in Fedora 8 and the same database with which there were
performance problems when migrating to 8.3. The migration was not
done. The thread was this:
http://archives.postgresql.org/pgsql-general/2008-01/msg00508.php
The database performance slowly degrades and after two weeks i issue a
reindex on the database and the performance gets back to normal. I
have been doing this for ages and i don't remember in which version
this degradation behavior appeared, perhaps 8.0 I'm not sure.
Every two or three months the reindex gets slow even for small tables
then i reboot the machine and it gets back to normal reindex
performance. This happened in more than one version of Fedora and
Postgresql and in two different production machines with very
different configurations.
Now what is happening is that reindex does not finish even with a
small 6,500 rows table and after a reboot. In top there is no CPU or
memory usage by postmaster and vmstat shows no disk activity.
I recently had a crash during a bulk insert when i updated to the
2.6.24.3-12.fc8 kernel. I rebooted and Postgresql did its thing to
recover and everything was working. I made the previous kernel,
2.6.23.15-137.fc8, the default in grub.conf and rebooted.
How do I know if the database is corrupted? Since there is no explain
for reindex or vacuum i don't know what is happening. How to debug?
Regards, Clodoaldo Pinto Neto
