Обсуждение: page x is unitialized

Is this message the sign of trouble, or is it harmless?

Michael Guerin <guerin@rentec.com> writes:
> Is this message the sign of trouble, or is it harmless?

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.)
If you haven't had any crashes lately then I'd wonder about whether your
disk is dropping data.

            regards, tom lane

Tom Lane wrote:

>Michael Guerin <guerin@rentec.com> writes:
>
>
>>Is this message the sign of trouble, or is it harmless?
>>
>>
>
>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.)
>If you haven't had any crashes lately then I'd wonder about whether your
>disk is dropping data.
>
>            regards, tom lane
>
>
Could this lead to the invalid page header problems that I've been
having lately?

Michael Guerin <guerin@rentec.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