Re: pg_xlog is getting bigger

Adrian Klaver <adrian.klaver@gmail.com> writes:
> On 12/19/2012 01:16 PM, Kevin Grittner wrote:
>> AI Rumman wrote:
>>> checkpoint_segments | 300

> Aah, just noticed the above.

I saw that too, but it doesn't seem to explain 20GB worth of pg_xlog.
The fine manual mentions somewhere that we can keep up to twice the
specified number of xlog segments, so one would expect at most 16MB
* 600 or about 9.6GB in pg_xlog.  (And this is assuming that
checkpoint_timeout is large enough to let checkpoint_segments be the
controlling factor.)  So there's something else going on.

            regards, tom lane