Adam Haberlach wrote:
>
> I've got a system in which a row is being created by one process,
> a notify is sent, and another process finds and then deletes the
> row. This whole interchange seems to be taking about a second, which
> seems oddly slow.
If you do it enough times without vacuum, the table grows quite big,
even
though it may look empty...
> As far as I know, I have the fsync-on-anything
> turned off (*).
>
> I'm mainly wondering if I should index/not index the key of the
> table, and if I should blame the hardware or Postgres?
Index would very likely help. (BTW, the best way to find out is to try
it -
you can always drop it later ;)
> The hard
> drive on the R&D server is pretty slow, although the CPU is pretty
> dang fast. I could also blame the notification system but that
> shouldn't be a problem, right?
How do you read the notify ?
----------
Hannu