"RESTORE VOLUME" process is slow

[msteffens]

I'm restoring a damaged Primary Pool tape volume from a Copy Storage Pool (90 tapes needed). There are 360,000 files that need to be restored. The process has been running for almost 24 hours now and it has only restored 280,000 files. I'm wondering if anyone else has experienced this process being slow or if there is any "show" commands that might show what's holding this process up?

Any ideas would be great. It's painful to watch!

Thanks for any feedback!

Marty

 

[moon-buddy]

Maybe normal given how fragmented (files scattered allover the 90 tapes) the files are. This should be faster with collocation (assuming that we are talking on restoring a group of nodes only) enabled.

"q proc" should give you an idea about wait times.

How many drives do you have? you can increase throughput by firing up more restore sessions.

 

[GregE]

As moon-buddy says, files scattered all over is the reason for this. Lots of mounting, dismounting, searching, etc.

Run this to see some basics.

select process_num as PROC_NUM, process, current_timestamp-start_time as ELAPSED_TIME,(cast(files_processed as decimal(29,2)) / cast((current_timestamp-start_time)seconds as decimal(29,2)))as "FILES/SEC", (cast(bytes_processed/1024 as decimal(29,2)) / cast((current_timestamp-start_time)seconds as decimal(29,2))) as "KB/sec",cast(BYTES_PROCESSED/1024/1024/1024 as integer) as GB from processes

 

[msteffens]

Thanks for the replies. I was under the impression that it would just reload the files from the tapes in order of the tapes that it produced in the "preview". That's not the case.

Greg... Here's the output from your query:

PROC_NUM PROCESS ELAPSED_TIME FILES/SEC KB/sec GB
13129 Restore Volume 1 04:43:14.000000 3.11 405.37 39
13318 Migration 0 01:11:30.000000 7.41 38617.11 157
13319 Migration 0 01:11:30.000000 16.42 39073.94 159

Definitely an eye opener for me on how my Copy storage pools are defined!!

Thanks for the help. I appreciate it!