RMAN: backup archivelog all, eats ALL processors!

Hi,

Although this might be a less than frequent real scenario in my particular case, I'm very worried about this behavior:

I'm testing a 9i (9.2.0) database in a Dell 8450 with Redhat 7.3, 6 processors and 4GB of physical memory. The database is in archivelog mode, automatic archiving is enabled (arc0, arc1). It is going to be used to store GIS data. After loading a bunch of tables (representing a huge map and it's features), archived logs are placed in a predefined location. After repeating the process a couple times, the directory containing archived logs has 2.2GB worth of inactive redo log records.

In RMAN, I configured a channel pointing to a different physical disk on the same system. Then, (assuming I need to backup this) I issue: backup archivelog all delete input;

The command runs normally for about one or two minutes. After that, all 6 processors get completely busy, and the systems becomes barely responsive. IO activity decreases to almost nothing, and this lasts at least two hours (I have never waited more than that for it to complete). TOP says it is Oracle who has the CPU resources. Even after a shutdown abort, the oracle process goes on as defunct, and the problem remains!

Any clues on what can be wrong??

Thank you.
Ivan. Received on Mon Nov 18 2002 - 13:26:00 CST