This content has been marked as final. Show 4 replies
Is it a full backup that you are taking? For 23TB, it would be a lot for RMAN to do so probably, what is appearing as hanging is just the time taken by RMAN to start the backup job.
You can collect the diagnostic to check what's going on behind the scenes using following metalink note.
*Troubleshooting RMAN Performance or Hang Issues [ID 815857.1]*
Also, How many total number of CPU cores you have on you machine? Oracle will not spawn channels more than our total number of cores at one single point of time. So in your case it might be possible that your all cores are active backing up datafiles and just waiting for them to finish before going any further.
I will keep an eye on it further.
I guess also the goal is to keep backup time completions to under 3 days, or even 2 days or even less, so I am tuning this as I go along.
Keep in mind, I WAS using MAXPIECESIZE and now I removed this,...I am not sure if this has anything to do with it.
Also, when it reached the CHANNEL 99, it just stayed there for hours..which was unusual...
100 channels seems like overkill. Do you really want 100 separate processes concurrently doing I/O ? Where are the backups being written to ? (e.g. if tape, then you really need very many tape drives, if disk how are you spreading the write I/O of those 100 channels ?)
Hemant K Chitale