Query inactive files - delete specific files

[xefil]

Hi to all!



I've a TSM Server 5.3 and many clientnodes making backup on a STGPOOL. This STGPOOL has more or less 100 volumes (disk).

Now the stgpool is near to be full. Is there a way to:



1. Query inactive files of a specific node and find out for example dump or big files not important?



2. Query content of filespaces without to know where volume has the files stored? with 100 volumes it's hard to query every volume.



3. Delete specific files, not filespaces?



Thank's in advice!



Simon

[may]

1- has PJ said in a post some sep down, you can a "exoprt node filedata=backup preview=yes"

then "exoprt node filedata=backupActive preview=yes" to calculate the diffrence

==> it will you give the exact amount and volume of Inactives files store for that node.



2. dont really understand your need,



3. since 5.3.x cient version, you have the "Delete backup" command that you can use from the client.

you can use it from gui (i 'll recoomand to select all object in the option, to be sur to delete the directry srtucture)

[PJ]

You should also be aware that on a diskpool space only ever reclaims, when an entire aggegate expires. So rather than deleting data you should probably first take a look at how much capacity you're currently wasting and (if that is significant) reclaim the space by reconstructing aggregates through a squential pool.



Cheers

PJ

[chad_small]

PJ



There is a better way to cleanup disk pools. Use the following command to clean up empty space in aggregates. Update the disk pool volume you want to clean up to read only then use



MOVE DATA <VOLUME NAME> RECONSTRUCT=YES



This will move the data to other volumes in the storage pool while cleaning up the empty space. As for deleting specific files, it is possible but USE AT YOUR OWN RISK! You need to use the following command to verify the files are ones you can do without first:



select object_id, backup_date, hl_name, ll_name from backups where node _name='<NODE_NAME>' and deactivate_date<'2006-10-28 00:00:00'



NOTE: PUT IN THE DATE THAT MEETS YOUR RETENTION REQUIREMENTS!!!



Then output this one to a file and remove the header:



select object_id from backups where node_name='<NODE_NAME>' and deactivate_date<'2006-10-28 00:00:00'





Then run it through the following unix shell script to process the file deletions:





#!/bin/ksh



while read LINE

do

dsmadmc -id=<admin ID> -pa=<admin pw> delete object 0 $LINE

echo Object $LINE Deleted

done < $1

[PJ]

Chad,

my understanding is, that reconstruction is and can only be supported with sequential data. If thats untrue I scripted some 500 lines perl for nothing (and would still be very happy in deed ).



Cheers

PJ

[chad_small]

PJ - You are correct! I misread the options parameters. Best bet would be to start using a file devclass since its considered sequential then you could do the cleanup of aggregates.