Ray Stell <stellr@cns.vt.edu> writes:
> On Wed, Feb 02, 2011 at 06:32:22PM -0500, Tom Lane wrote:
>> "unexpected pageaddr" is just one of the standard tests for detecting
>> end of WAL, so I don't think this is anything to be surprised about.
>> It looks to me like the standby applied all the WAL it had and then
>> connected to the master for more.
> thanks, Tom. I suppose it's the order that trips me up:
> starts at 65...
> reached a 66...
> unexpected at 62...
Well, the point of the "unexpected pageaddr" message is exactly that the
WAL location it found in the next file wasn't the expected next location
in sequence (which would have to be something larger than 0/66000000).
This indicates that the next file isn't really new WAL, but a segment
file that was recycled (by renaming it to a higher file number) but not
yet overwritten with new data.
regards, tom lane