Greg Smith wrote:
My guess is that there's something wrong with your config such that
writes followed by fsync are taking longer than they should. When I
see "sync=0.640 s" into a SAN where that sync operation should be near
instant, I'd be looking for issues in the ZFS intent log setup, how
much data you've setup ZFS to write cache, and the write cache policy
on the SAN hardware. There's something weird going on here, that sync
should be near instant on your hardware with such a small write volume.
Unfortunately, you may have to take the server down to find out
exactly what's going on, which doesn't sound practical for your
environment. (And people wonder why I have near religious zeal about
testing disk hardware before systems go into production)
there's a further complication. this system is deep in southeast asia
on a rather slow and saturated network connection such that trying to
figure out stuff about it is challenging at best. I know the database
has a lot of write volume overall, and its only one of several databases
running in different zones on the server. I know nothing about the SAN,
I suspect its a EMC Symmetrix of some sort. Probably a generation or
two behind latest. The operations people are used to running large
oracle databases.
--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
