Legato Networker Server Migration

[iamwhitey]

Currently we are running Networker on a Solaris Server. Well that server is getting up in years and needs replaced. We decided to bail on Sun and migrate to a kickass dell server running Red Hat.

I realize cross platform migrations are not supported by Legato. Has anyone migrated from Solaris to Linux? If so what help can you give me?

My current plan was to copy over the indexes from the old server to the new server. If i leave the host and ip all the same I would think it would work.

Any help will be greatly appreciated.

Thanks

John

 

[Chacalinc]

I don't know what could happen with the indexes, I guess there is no problem, but never tryed it. The only issue would be the licenses (host id will be differents).

I would try to save the bootstrap and then recover it in the new server.

Cheers.

Chacal, Inc.

 

[605]

There will in fact be a problem - this is due to the fact how microprocessors will calculate internally - the 'little endian' or the 'big endian' way which means 'least or most significant bit/byte first'.

Years ago i briefly tested this and could even successfully transfer indexes between Windows and Linux. However - Intel and Sparc, this will not work for sure.

 

[iamwhitey]

Yes i guess that would be a problem. So I guess the best solution would be to put in the new server and reindex the server. am i correct? How long does a reindex take on say 350 clients?

 

[605]

I do not think so. NetWorker's recover mechanisms should do this properly and automatically. So a standard DR recovery with "mmrecov" and "nsrck -L7" should do the job and recover the metatdata. If this is a general problem i would expect a lot of complaints on this alias.

What you just can not do is to FTP the metadata among the servers.

 

[iamwhitey]

well the time came and we were able to bring the system backup over on the linux system using the mmrecov command with no problems.

Currently the only problem is Red Hat working with the drives. but the application side everything was smooth.

 

[adecnire]

You may have block size problems after you had migrated to red hat. Im currently working on migration. From solaris to suse. If you cant read the tapes from red hat, try changing the block size to variable. You can do it by;

mt -f <device name> setblk 0

Before running this command the device should be loaded with a tape. Try loading the tapes with

nsrjb -nlv -S(slot) -f(device name)

good luck