Michael Guerin <guerin(at)rentec(dot)com> writes:
> Tom Lane wrote:
>> It's not good. The least bad explanation is that you had a system crash
>> recently that occurred between extending a table by one page and filling
>> that page with useful data. (The page is filled with zeroes on creation.)
> Could this lead to the invalid page header problems that I've been
> having lately?
No, more the other way around. IIRC, "page n is uninitialized" is
VACUUM's report that it is fixing an invalid page header.
regards, tom lane