Обсуждение: Errors with pg_dump

I would like to move some data from an older installation of PostgreSQL to
a newer.  When doing
"pg_dump persondb > db.out" I get the following error message:

"dumpSequence(person_sek): 0 (!=1) tuples returned by SELECT"

The "person_sek" is a sequence in the database.

The version of PostgreSQL in question is 6.3.2, running on RedHat Linux
5.1/5.2.  The actual database (persondb) seems to be running fine in all
other respects.  It can be queried with psql, and is used as the backend
for a set of web pages.

Help, anybody?

Asbjørn Sæbø

>
> I would like to move some data from an older installation of PostgreSQL to
> a newer.  When doing
> "pg_dump persondb > db.out" I get the following error message:
>
> "dumpSequence(person_sek): 0 (!=1) tuples returned by SELECT"
>
> The "person_sek" is a sequence in the database.
>

I believe sequences are implemented as a separate with one row that contains
the sequence parameters and state.  It looks like somehow that one row has
been deleted and pg_dump expects it to be there.

Is this sequence being used by your program and is it functioning correctly?
Can you do a select nextval('person_sek')?

If you need the sequence and you know what its current value and other
parameters should be then I suggest droping and recreating the sequence.

asbjs@stud.ntnu.no writes:
> I would like to move some data from an older installation of PostgreSQL to
> a newer.  When doing
> "pg_dump persondb > db.out" I get the following error message:
> "dumpSequence(person_sek): 0 (!=1) tuples returned by SELECT"
> The "person_sek" is a sequence in the database.
> The version of PostgreSQL in question is 6.3.2,

Hmm.  Does the sequence still work (can you do SELECT nextval('person_sek'))?

Not sure why the dump attempt would be failing, and 6.3.2 is far enough
back that digging for bugs in it isn't very appealing.  I'd suggest just
looking for a work-around instead of a real solution.

You could probably just drop and recreate the sequence before running
pg_dump, being careful to set the new sequence's initial value to
whatever its current value is.

            regards, tom lane