Undo tablespace keeps on growing

Question

We have a 3rd party application running on our oracle(10.2.0.4) database running on 64bit solaris(sparc) machine.
Since few days my undo started going up. I increased the size from 2g to 5g. Still it is at more then 90%levels.
I ran this query to see which session is using maximum undo
SELECT a.sid, b.name, a.value
FROM v$sesstat a, v$statname b
WHERE a.statistic# = b.statistic#
AND a.statistic# = 176
ORDER BY a.value DESC
From this i found the session which is using max undo.
When i query this sid from v$session i see this is inactive
my undo retention is set to 9000 and undo management is auto
how can i check if i have expired undo blocks that are not being used

Answer 1

>
Since few days my undo started going up. I increased the size from 2g to 5g. Still it is at more then 90%levels.
I ran this query to see which session is using maximum undo
SELECT a.sid, b.name, a.value
FROM v$sesstat a, v$statname b
WHERE a.statistic# = b.statistic#
AND a.statistic# = 176
ORDER BY a.value DESC
From this i found the session which is using max undo.
When i query this sid from v$session i see this is inactive
my undo retention is set to 9000 and undo management is auto
>
With an UNDO_RETENTION of 2.5 hours set, I am not surprised that you use up to (or even more than) 5g space in the UNDO tablespace! I would even call that moderate.
What is your concern? If you don't have the space for your UNDO tablespace, lower UNDO_RETENTION accordingly. Notice that UNDO_RETENTION is the wish to preserve before images for that long time even if their transaction did already commit. Why have you set it to that (relatively high) value of 2.5 hours?
Kind regards
Uwe
http://uhesse.wordpress.com