Block size setting on a Solaris server

Question

Hi, I have now a NW server v7.2.1 in production on a HPUX v11.11 machine. It uses a IBM3494 library with 3592 drives type. Due to HBA settings, I suppose, the networker server uses a block size on its tape 256K in size, while networker has per default a blocksize 384K in size, for the same 3592 model. The datazone has 20 storage nodes, both HP and Sun, using each its own blocking factor ( tipically 256K HP, 384K Solaris ). During one year in production, no problem to backup/restore. Now I have to migrate the server on a Solaris v9 platform, Networker v7.2.1. I've seen that the HBA driver, working with networker , uses a blocksize 384K in size. I obviously have to use all the tapes I've already used on HPUX, to recover data from or to append new backup to, and therefore I have absolutely to be sure that there are no misunderstandings reading tapes labeled/written on HP from the new Solaris server. My idea is to set NSR_DEV_BLOCK_SIZE_3592 variable to 256 on the new Solaris server, adding it to the networker start script. I think also that this variable's change will affect only the networker server, and not the storage nodes ( thay'll continue to use their own blocking factor ). I'll do the right thing setting up this variable on the solaris server ? It's the right way to be sure that all the tapes written on the HP platform will be readable/rewritable on the Solaris one ?? And what about the storage nodes ??? Thanks a lot for any suggestion Pierpa

ble1 · Answer

If something is written with bigger block size than your new HW can support then you will have issue. You may wish to force it by setting device block size parameter in device properties, but still I have no idea if that could be applied to already written tapes. You will get errors, but I have no idea if they will be just errors and restore will proceed or it will fail. Try to see if you can get somehow HBA working with higher bs rate.

NetWorker

Was this post helpful?