Re: Avoid duplicated rows when restoring data from pg_dumpall ??

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Tom, thanks for your fast reply,

- As regards the duplicated rows, no, I don't get duplicated rows in all the tables stored in the database because
some tables have primary-keys (and/or UNIQUE) constraints. These constraints don't allow the restore process to
duplicate rows.  In fact, it is a kind of "solution" I've tried...(add an extra column with a primary-key or unique constraint, to this tables), and it "works". The restore process doesn't generate duplicated rows, because the constraint does not allow the insertion
of new duplicated data.  Anyway..it looks like a kind of 'poor solution'  :-)

- Ok, thanks for the info, I thought pg_dumpall would work as I desired even on non-empty clusters.
Of course...if there is no previous data, the restore process will never create duplicated rows.

- Yes, the restore process generates errors, because it tries to re-generate data structures that exist in the database server
at that moment. Even if I delete my own databases, some errors will appear (because I cannot delete the internal stuff of
of the server -> the 'postgres' database for example) . Those errors, could be ignored in most cases I think, but perhaps create a kind of "bad feeling" about the result of the restore process,  or can "hide" other more important errors when you get a huge ammount of info on the screen either.

- I forgot to mention it. I'm using PostgreSQL 8.3 on Windows xp.

Thanks a lot for your advices,

Regards, Pablo

2009/8/24 Tom Lane <tgl@xxxxxxxxxxxxx>
Pablo Alonso-Villaverde Roza <pavroza@xxxxxxxxx> writes:
> I'm getting duplicated rows in some of my tables when I try to restore data
> from a dump file generated with 'pg_dumpall'.

Probably all of them, actually ...

> The only "way" I have found to solve this problem is deleting my database
> before restoring the data, so everything is 're-created' in the restore
> process without generating duplicated rows.

A pg_dumpall script expects to be restored into an empty cluster.  This
is not a bug.

> I thought, that the "-c" flag on pg_dumpall would force a DROP of any
> previous data structures on the server but...it looks like it doesn't work
> as I expected and ...when I restore data I get duplicated rows.

The -c flag should cause the script to DROP all your databases first.
But that switch has been known to have bugs in the past, and in any case
it could fail if there are other sessions connected to those databases.
Did you pay attention to whether the restore reported any errors?
What PG version is this?

                       regards, tom lane


[Index of Archives]     [KVM ARM]     [KVM ia64]     [KVM ppc]     [Virtualization Tools]     [Spice Development]     [Libvirt]     [Libvirt Users]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite Questions]     [Linux Kernel]     [Linux SCSI]     [XFree86]

  Powered by Linux