Would something like this help?
select node_name, sum(num_files) as "# FILES",
sum(physical_mb) as "MB (PHYSICAL)",
sum(logical_mb) as "MB (LOGICAL)"
from occupancy
group by node_name
order by "MB (PHYSICAL)" desc
This will show you the total number of files, physical MB,
and logical MB used by each node, sorted by physical MB in
descending order (so the big hitters appear first). You can
alternatively sort by "# FILES" (or any other field) if you
wish.
Monitoring this on a daily basis may show you who is growing.
Sample output (I manually reduced the field widths here to
help prevent line wrapping of this note):
NODE_NAME # FILES MB (PHYSICAL) MB (LOGICAL)
--------- ----------- ------------- ------------
STORMAN 669 50.59 50.59
STORMAN 669 50.59 50.59
DECALPHA 207 35.54 35.54
KETURY 23 16.65 16.65
ARIZONA 1 7.74 7.74
PTF3 7 0.93 0.93
WIN98 53 0.26 0.26
MOCKV2 24 0.13 0.13
MOCK 300 0.08 0.08
Regards,
Andy
Andy Raibeck
IBM Storage Systems Division
ADSM Client Development
e-mail: storman AT us.ibm DOT com
Well, I don't know about all files, but if you want to know about the tapes
per node, try this select:
select node_name,count(*) from volumeusage group by node_name
Mark Mapes
PG&E
> ----------
> From: Bob Brazner[SMTP:Bob.Brazner AT JCI DOT COM]
> Reply To: ADSM: Dist Stor Manager
> Sent: Tuesday, September 15, 1998 10:14 AM
> To: ADSM-L AT VM.MARIST DOT EDU
> Subject: Need help controlling explosive growth
>
> During our daily DRM process, we backup the storage pool to tape. In
> the last two weeks, we've nearly doubled the number of tapes we
> generate. How can we find out who are the big users? I see no total
> usage stats maintained at the filespace level (too bad). How do I write
> QUERY/SELECT on all files (which number in the hundreds of thousands),
> grouped by node, without it taking forever to compute? Am I better off
> dumping the query output to disk and then using something like SAS to
> process? Any ideas anyone?
|
