Problem with sequences on a reload of a pg_dump file

Started by Sean Shannyalmost 22 years ago4 messagesgeneral
Jump to latest
#1Sean Shanny
shannyconsulting@earthlink.net

To all,

We dumped our instance from a 7.4.2 instance compiled with a block size
of 32k running on OSX 10.3.3 server.. We loaded this dump on Fedora
2.6.5-1.358smp running postgresql 7.4.2 compiled with the standard 8k
blocks size.

My problem is that all the sequences were reset to start a 1. :-( Did
I do something wrong as this has never happened before? Also, can I
simply figure out what the next value should be and manually update the
last_value column in each affected sequence?

Thanks.

--sean

#2Tom Lane
tgl@sss.pgh.pa.us
In reply to: Sean Shanny (#1)
Re: Problem with sequences on a reload of a pg_dump file

Sean Shanny <shannyconsulting@earthlink.net> writes:

My problem is that all the sequences were reset to start a 1. :-( Did
I do something wrong as this has never happened before?

Did the reload actually complete, or did it fail partway through?

Also, can I
simply figure out what the next value should be and manually update the
last_value column in each affected sequence?

No, but you could use setval().

regards, tom lane

#3Sean Shanny
shannyconsulting@earthlink.net
In reply to: Tom Lane (#2)
Re: Problem with sequences on a reload of a pg_dump file

Tom,

It looks like I probably caused this. :-( I was not paying attention
and rebooted the machine while it was doing the index building. :-(
This is about an 1/2 TB warehouse instance so I am going to look at the
last several thousand lines of the dump file to see if I can replay what
I need to correct the situation. Not a good day.

--sean

Tom Lane wrote:

Show quoted text

Sean Shanny <shannyconsulting@earthlink.net> writes:

My problem is that all the sequences were reset to start a 1. :-( Did
I do something wrong as this has never happened before?

Did the reload actually complete, or did it fail partway through?

Also, can I
simply figure out what the next value should be and manually update the
last_value column in each affected sequence?

No, but you could use setval().

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match

#4Yannick LOUVET
y.louvet@atolcd.com
In reply to: Sean Shanny (#1)
Re: Problem with sequences on a reload of a pg_dump file

non non c plus possible
va falloir penser � arreter tout de m�me !!!!

Sean Shanny wrote:

Show quoted text

To all,

We dumped our instance from a 7.4.2 instance compiled with a block
size of 32k running on OSX 10.3.3 server.. We loaded this dump on
Fedora 2.6.5-1.358smp running postgresql 7.4.2 compiled with the
standard 8k blocks size.

My problem is that all the sequences were reset to start a 1. :-(
Did I do something wrong as this has never happened before? Also,
can I simply figure out what the next value should be and manually
update the last_value column in each affected sequence?

Thanks.

--sean

---------------------------(end of broadcast)---------------------------
TIP 8: explain analyze is your friend