Re: postgres 8.2.9 transaction id wraparound failure

I tried just vacuuming the "postgres" database first. No luck. Was
still unable to startup the server normally.

Here's is a snipped from the query you suggested.... All the databases
on this machine look similar.....

backend> SELECT datname, age(datfrozenxid) FROM pg_database ORDER BY
age(datfrozenxid) DESC
2009-04-20 14:27:52.250 EDT [10097] [] WARNING: database "postgres"
must be vacuumed within 981218 transactions
2009-04-20 14:27:52.250 EDT [10097] [] HINT: To avoid a database
shutdown, execute a full-database VACUUM in "postgres".
1: datname (typeid = 19, len = 64, typmod = -1, byval = f)
2: age (typeid = 23, len = 4, typmod = -1, byval = t)
----
1: datname = "postgres" (typeid = 19, len = 64, typmod = -1,
byval = f)
2: age = "2146502429" (typeid = 23, len = 4, typmod = -1, byval = t)
----
1: datname = "ange" (typeid = 19, len = 64, typmod = -1, byval = f)
2: age = "2146502429" (typeid = 23, len = 4, typmod = -1, byval = t)
----

Alvaro Herrera wrote:
> Maria L. Wilson wrote:
>
>> I was afraid you'd say that - any quick way to vacuum all 315 databases
>> on this machine?
>>
>
> Actually you only need to vacuum the database being complained about --
> "postgres" in this case. You can issue
> SELECT datname, age(datfrozenxid) FROM pg_database ORDER BY age(datfrozenxid) DESC
> to find whether any other database is going to need vacuuming as well.
>
>