Обсуждение: dump sizes changed...

We recently migrated to a new server after having had the old one
running continously for over a year.  The first night that we did a
backup, the file produced by pg_dumpall was noticably smaller than the
one we got from the old server prior to the switchover (2.5 GB versus
3.1 GB).

Is this something we should be alarmed about, or could this be the
normal result of dumping and reloading a large database?

Thanks for any who can advise.
--

Thank you,

Lewis Kapell
Computer Operations
Seton Home Study School

Lewis Kapell <lkapell@setonhome.org> writes:
> We recently migrated to a new server after having had the old one
> running continously for over a year.  The first night that we did a
> backup, the file produced by pg_dumpall was noticably smaller than the
> one we got from the old server prior to the switchover (2.5 GB versus
> 3.1 GB).

> Is this something we should be alarmed about,

It requires investigation at least.  Are these the same PG version, or
different ones?

We modified pg_dump awhile ago to be less verbose, but I think it was
only shortening the "comments" a bit; your result sounds too different
to be explained by that.

            regards, tom lane

Sorry, I should have given the versions.  The old server was running
8.0.8, the new server is running 8.2.1.

Thank you,

Lewis Kapell
Computer Operations
Seton Home Study School


Tom Lane wrote:
> Lewis Kapell <lkapell@setonhome.org> writes:
>> We recently migrated to a new server after having had the old one
>> running continously for over a year.  The first night that we did a
>> backup, the file produced by pg_dumpall was noticably smaller than the
>> one we got from the old server prior to the switchover (2.5 GB versus
>> 3.1 GB).
>
>> Is this something we should be alarmed about,
>
> It requires investigation at least.  Are these the same PG version, or
> different ones?
>
> We modified pg_dump awhile ago to be less verbose, but I think it was
> only shortening the "comments" a bit; your result sounds too different
> to be explained by that.
>
>             regards, tom lane