"Joshua D. Drake" <jd@commandprompt.com> writes:
> Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> IIRC we were speculating that data was being written
>> in a pattern that required a lot of seeking thus ruining throughput,
>> but we didn't have any hard evidence of that. Did you do the
>> strace'ing I suggested?
> Yes, I asked if you wanted counts or the whole output. You didn't
> answer :). I provided the counts.
Uh, sorry. Counts are useless here, we need to see the sequence of
write locations to find out if there's a lot of nonconsecutive
writes happening. BTW, the strace had better run across the whole
PG process tree --- it's quite possible that there's some interaction
between the bgwriter and the backend doing COPY, for instance.
regards, tom lane