Thanks.

George

Dave Mussulman wrote:

On Tue, Nov 08, 2005 at 11:22:14AM -0500, George Sinclair wrote:

Hi,

We have an older NetWorker release (6.1.1) running on Solaris 2.8, backing up mostly Linux clients, and we have a Linux storage node server running the same release. We're running a Quantum P1000 SDLT 1 tape library and a STK L80 LTO 1 tape library on the storage node and an older DLT 7000 tape library on the primary server. We can no longer continue to use the same physical primary server box as it's no longer under hardware support from Sun, and we're way behind on our NetWorker release which I suspect is fraught with bugs. We've seen sporadic problems with block size error messages during backups, etc. Legato support has basically told us that there were a lot of issues that were corrected in later releases, and they can't really help further us until we upgrade. We have the option of moving the primary server to a newer Sun Ultra 4 box running Solaris 2.9 or a Linux box (RedHat ES release 3) but either way it will be a fresh install. we back up directly to the tape drives using the LAN, no SAN involved.

o What is the latest version of the NetWorker software that we *should* use? Does anyone have any recommendations or advice on the install and or pit falls to avoid? Is it reasonably straight forward?

o We don't have the latest media kit, but we do have support. Not sure why they've not sent us anything, so not sure which media kit to ask for? I've never seen any checksums or signatures for anything they have on line so was thinking the media kit would be safer.

o I've heard there are some security patches? Is this a separate step or included in the release?

I would probably go with 7.2.1, as that has the latest security fixes
that I'm aware of.  There used to be some 7.2 hesitation on the list,
but it looks like enough sites are using 7.2.1 that things are okay.
I've never installed off the media kit; always off the website.

o If we set up a temporary licensed server on the new host, is it possible to transfer a few of the existing clients over, test some backups/recovers but continue to backup the other hosts on the current (older) server until we can get everything moved? The reason I ask is because I doubt we can complete all the testing and transfer of indexes in one day. Might require a week or more. Not sure if Legato will allow two copies of the media database to exist on two separate servers since each would be updated nightly. At some point everything would need to be merged. Can this be done, and if so, how?

Testing is one thing, cutting over is another.  You should be able to
setup another copy of the server somewhere else and get a 45-day grace
license.  However, once you've done the testing, I don't think there's
any way to merge a test system into a new system.  At some point, there
needs to be a cutover from the old to new system.  The licensing part is
easy; that can be done via email to get new codes for the new server.

o Does anyone know if NetWorker depends on the endianess of the machine? For example, since Linux is little endian, and Solaris is big endian then if I copy a client index over to a Linux box will it have a problem interpreting the data? Obviously, the client indexes are non-sparse, and md5 checksums can be used to validate the transfer, but I'm concerned about any potential endian problems. I have no way of knowing if NetWorker relies on the endianness of the OS or not. If this is an issue, and we move to Linux, we'd need to re-run fulls and start all the indexes over again from scratch. That would be very time consuming with 60+ clients. They say they do not support transferring to different OS, but they don't say why. Is this the reason? Should we just play it safe and stick to Solaris?

The trick is the Networker tape format is endianless; you can restore
to/from different OSes without running into that problem.  When I
migrated from Solaris to Linux, I did a bootstrap backup to tape, and
then did the normal DR recovery for that on my new Linux machine.  Any
indexes I moved over were from nsck -L7 restores from tape.  I did this
on a few machines to spot test that it worked.  I didn't bother
restoring the indexes for all of the clients; I did a full backup
relatively soon after the transition.  If I needed a browseable restore
from some of these clients, I knew I could go back to the older indexes.

Somewhere in the transition, the one hitch I ran into was being able to
do those index restores.  I had to put a /nsr/debug/no_striped_recover
file in place to be able to restore them.  I never got a good reason why
this failed after the transition, but putting the file there resolved
it.  I hope after I upgrade to 7.2.1 to try removing that file and see
if whatever was causing a problem was resolved.

I also remember, in one of a few conversations with EMC support after
the transition, learning about a better way to transition/rename the
server ... but unfortunately, I didn't write it down.  EMC doesn't
support moving across OSes, but they do support renaming and moving
machines between architectures, so feel free to ask them if you have
questions with any of those parts.

Dave



I recently migrated from Solaris to Linux, along with renaming the

backups server.

To sign off this list, send email to listserv AT listserv.temple DOT edu and type 
"signoff networker" in the
body of the email. Please write to networker-request AT listserv.temple DOT edu 
if you have any problems
wit this list. You can access the archives at 
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

To sign off this list, send email to listserv AT listserv.temple DOT edu and type 
"signoff networker" in the
body of the email. Please write to networker-request AT listserv.temple DOT edu 
if you have any problems
wit this list. You can access the archives at 
http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER