Bruce Momjian <[EMAIL PROTECTED]> writes:
> Tom Lane wrote:
>> ... The question stands though: why isn't it
>> appropriate to warn of overly-frequently-issued manual checkpoints?

> ... the warning is for cases when you are filling up the WAL logs too
> quickly and checkpoints are happening too frequently.  If a user is
> doing checkpoints, it isn't anything increasing the checkpoint segments
> is going to help.

No, I think the warning is for when checkpoints are happening too
frequently, period.  An overly small checkpoint_segments setting
is one possible cause of that, but the performance penalty from
too many checkpoints is just as bad no matter what's causing it.
(Remember that a checkpoint not only forces I/O in itself, but
significantly increases subsequent WAL traffic because of needing
to dump whole page images into WAL.)

How do you feel about improving the signaling mechanism but using
it just to vary the HINT?

LOG: checkpoints are occurring too frequently (nn seconds apart)
HINT: Consider increasing the configuration parameter "checkpoint_segments".

LOG: checkpoints are occurring too frequently (nn seconds apart)
HINT: Issuing explicit CHECKPOINTs so often is really expensive.

and so on.

                        regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
      joining column's datatypes do not match

Reply via email to