It's time to rant, briefly, about the ADSM documentation.

We are in the process of transitioning to a V3 server running
under AIX. As a test, we decided to export a single 32GB
filespace for one of our large clients. The 'export node' command
for this filespace took roughly 18 hours, and the data occupied
seven (7) 3590 cartridges.

After studying the docs for the 'import node' command, and
a few unsuccessful attempts at the syntax, we got the 'import
node' running. Not being discouraged when the server hung, we
halted and restarted and tried again. The second time the 'import
node' was running beautifully. It looked like it was going to
take approximately six (6) hourse to complete, well within our
window, and in time to start an incrmental backup of that
filespace to the new server last night.

Unfortunately, when I checked on the process last night, I
discovered that it had terminated unsuccessfully. It turns out
that when you specify the "-volumenames" parameter to the 'import
node' command, you have to put them in the order they were
created. Wouldn't you know it? I transposed the last two volumes,
so the process died when it wanted tape #6 and it got tape #7.

There is absolutely no mention in the documentation of a
requirement that the volumenames be given in any particular
sequence. These two processes combined to occupy 13 tapes, and
almost 24 hours of tape drive and CPU resources, all for naught.

Needless to say I was *quite* disappointed with ADSM last night.

"If I could dot the 'i' in a Michigan     Thomas A. La Porte
 game and the good lord came to take me   DBA, Feature Animation
 the next day ... at least I could        DreamWorks SKG
 die happy." - Beano Cook, ESPN           <tlaporte AT anim.dreamworks DOT com>