Here is a simple one I created. It counts all the files by node. The one
downside is that if the file is replicated in a copypool it counts it
twice. Someone might be able to help write around that if needed:
tsm: TSMSERVX>q script FILE_COUNT f=d
Name: FILE_COUNT
Line Number: Description
Command: # of files per node
Last Update by (administrator): BBULLOCK
Last Update Date/Time: 02/20/01 09:17:31
Name: FILE_COUNT
Line Number: 1
Command: select node_name, sum(num_files) as
"number of files" from
occupancy group by node_name order by
"number of files"
Last Update by (administrator): BBULLOCK
Last Update Date/Time: 02/20/01 09:17:31
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:[EMAIL PROTECTED] On Behalf Of
Nick Laflamme
Sent: Thursday, September 13, 2007 11:36 AM
To: ADSM-L@VM.MARIST.EDU
Subject: Object counts per node
We haven't done a good job of keeping balanced the number of nodes on
our farm of TSM servers, so we're looking at manually moving some of
them off of a server whose DB is close to full and can't be easily
extended. I can't find, though, any TSM queries that would show me how
many files or other objects a node owns. I can find information about
how large a filespace is and about how much storage space a node's data
takes, but not the number of objects adding up to that much space.
Any SQL queries I try to get me into that direction get me "this is a
large query" warning, which seem depressingly accurate based on how long
I wait for them before giving up.
TSM 5.3.4.2 on AIX, so we can't use any new commands from TSM 5.4 yet.
The crude way would be to guess that Windows or Mac clients have more
small objects and move those first, but if I can be just a wee bit more
scientific about this, I'd gladly make the effort.
Any suggestions?
Thanks,
Nick
