Zitat von Tom Lane <tgl@sss.pgh.pa.us>:
> A checkpoint pushes out all unwritten data since the last checkpoint.
> So yeah, it stands to reason that if you increase the time between
> checkpoints, each checkpoint will take longer. Whether this is really
> a problem is not clear --- the checkpoint is happening in background
> after all. Why should you care how long it takes?
>
> regards, tom lane
>
>
What I want to do, is to write a lot of data in the database over a longer
period (assume 1 hour or longer) with constant data rates - I am trying to use
Postgres to store measurement data.
When I only had a few checkpoint_segments (3 I think), I always got an
interruption when the database was checkpointing (what took about 3 seconds).
So I increased the checkpoint_segments and the pauses went away.
But now I think if I reach the segments limit again, I will get even bigger
pauses (of course at some point in the time a checkpoint has to be performed).
By the way, if you say, checkpointing is happening in the background, I don't
know what causes the pauses.
-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/