Re: transaction id wraparound

Interesting... age(xid) isn't documented anywhere.

No, vacuum shouldn't be generating a lot of xid's. My guess is that your
generating process actually does 2 transactions per row.

On Thu, Sep 28, 2006 at 11:16:24AM -0700, Sriram Dandapani wrote:
> When I run this query
>
>
>
> fwdb01=# select current_timestamp,datname,age(datfrozenxid) from
> pg_database;
>
> now | datname | age
>
> -------------------------------+-----------+------------
>
> 2006-09-28 18:04:24.489935+00 | postgres | 1087834006
>
> 2006-09-28 18:04:24.489935+00 | fwdb01 | 2039254861
>
> 2006-09-28 18:04:24.489935+00 | template1 | 2039253122
>
> 2006-09-28 18:04:24.489935+00 | template0 | 1542808250
>
> (4 rows)
>
>
>
> fwdb01=# select current_timestamp,datname,age(datfrozenxid) from
> pg_database;
>
> now | datname | age
>
> ------------------------------+-----------+------------
>
> 2006-09-28 18:10:45.64452+00 | postgres | 1088357075
>
> 2006-09-28 18:10:45.64452+00 | fwdb01 | 2039777930
>
> 2006-09-28 18:10:45.64452+00 | template1 | 2039776191
>
> 2006-09-28 18:10:45.64452+00 | template0 | 1543331319
>
>
>
>
>
> In approximately 6 minutes, the fwdb01 count has gone up by about 500K.
> I am generating about 250K rows for every 6 . I am also running
> vacuumdb.
>
>
>
> Does vacuumdb generate a lot of transactions that affects this counter.
>

--
Jim Nasby jim(at)nasby(dot)net
EnterpriseDB http://enterprisedb.com 512.569.9461 (cell)