Printable View
I have a target database 9.2.0.3 that is being backed up by rman (9.2.0.3). It is very slow. I have a 60G database and it is taking 5 hours.
When I log into rman to just do a crosscheck archivelog command it is taking over an hour. This seems to be the bottleneck.
I checked metalink and it said it could be due to a bug but has no workaround.
Has anyone else experienced this? Were you able to fix this?
Is it become slow recently or has it been allways slow?
What is the configuration? Directly to tape or to disk?
How many channels are you using?
Provide more info...:rolleyes:
It has been slow for a while.
I increased the channels to 3 from 1 recently but that did not do much. I am going to disk not tape.
what are your wait events?
The major wait is free buffer waits in the RMAN catalog database. I see it running this for a long time:
DELETE FROM ckp
WHERE dbinc_key = :b3
AND ckp_key NOT IN (:b2, :b1)
AND ckp_key NOT IN (
SELECT NVL (MAX (ckp_key), 0)
FROM ckp
WHERE dbinc_key = :b3
UNION
SELECT start_ckp_key
FROM tsatt
WHERE dbinc_key = :b3
UNION
SELECT NVL (end_ckp_key, 0)
FROM tsatt
WHERE dbinc_key = :b3
UNION
SELECT start_ckp_key
FROM dfatt
WHERE dbinc_key = :b3
UNION
SELECT NVL (end_ckp_key, 0)
FROM dfatt
WHERE dbinc_key = :b3)
In the target database I see waits on SQL Net Message from Client. I think it is waiting on the RMAN catalog.