No Query Restore Problems

[bot403]

Hello,



I am part of a large scale migration of data from TSM. My objective is to dump 100TB+ onto disk so that they can be archived into the new system. I am using a collection of custom perl scripts.



The goal is to get files off as quickly as possible so I am doing a no-query retrieve.



I am running into many problems doing a no-query retrieve of files. I want to use the filelist option so that we get files off efficiently, however I cannot specify a new destination on a per-file basis, only a destination directory. This is a problem because we have files which have the same name but are different. TSM will overwrite the same file several times when it does a restore.



I have tried the preservepath option but our filespaces are set up in such a way that the files will still be overwritten.

Ex.

fs = /fs1/dir1/file

fs = /fs2/dir1/file

fs = /fs3/dir1/file



In this case with "dsmc retrieve --preservepath=complete filelist=foo.txt /dump/"

I end up with 1 copy of /dump/dir1/file instead of 3.



Can anyone suggest what I should do?



Am I limited to pulling off files 1 at a time so I can assign them a temporary unique name?



I am open to any way to efficiently get our 100TB onto disk. Has anyone done something similiar?



I absolutely can not retrieve the files back to their original location and I am not concerned about what the files are named when they come off of TSM, only that they are retrieved and on disk.

 

[chad_small]

Unless I am mistaken there is no such thing as a NO QUERY RETRIEVE. Since ARCHIVED data is stored differently (packages due to the description name parameter) you have to select or state what you need specificly. You could add a timestamp to the files retrieved but then it could get tricky trying to identify them later. My question is why move them? If you have that much data in TSM just leave it and continue to upgrade the system as needed or move the TSM DB to a new server when the othe becomes to old. Seems dumb to even try with that much already archived.

 

[bot403]

Thanks for the reply,



First, I am but a lowly programmer so it is not my decision if the data stays or goes, it just has to move. We are migrating FROM TSM to a new system completely so we have to dump the data and then archive it again into the new system



Second, maybe I have my terminology mixed up since I am new to TSM. What I mean by "no query retrieve" is that I give it a filelist option so it will reorder the tapes efficiently when retrieving the data. Is this not the case? Is the filelist option no better than calling dsmc 1 file at a time with a retrieve command? If its not I can stop all this hoopla right now



Each file has a completely unique "filespace/path/filename" because all our files we are moving have a date as part of the filename so I do not see a problem just asking for it by filename, but the problem is just "path/filename" isnt unique across filespaces. Because of that TSM overwrites the files when we retrieve them even with the preservepath option.



I would really appreciate an answer so let me know if more examples would help clarify. We do have a unique setup so I could understand the confusion.