(no subject)

[Steve Cleveland]

> (Aside:  Do you know the command "pkill" in the more recent Solaris
> releases?  A single "pkill smbd" command can assist such firefighting.)

Thanks for the tip!  I was doing a funky foreach command, but pkill is
much better!

> Executive summary: run (don't walk) to your nearest Samba mirror and get
> release 2.2.3a .
>
> Detail:  Problems relating to oplocks, and to such smbd-avalanching have
> been noticeable in the the combination of Solaris and the evolving 2.2.x
> samba. By release 2.2.2 things were much better, but there were still a
> couple of significant remaining problems in precisely these areas.
>
> Jeremy has continued to do much work in this area.  Sometime after 2.2.2
> was released, he had one of those "Eureka!" moments (forgive poetic
> licence), and found and fixed the smbd-avalanching bug.  This fix was
> incorporated into Samba 2.2.3a .  So I strongly recommend that you plan a
> migration from 2.2.2 up 2.2.3a .
>
> Hope that helps.

I upgraded early this afternoon after another melt-down, and things seem
to be running much smoother.  The number of smbd processes is within a few
of the number of users on the system (using utmp).  Before it was always
off by 30-40.  Plus, the machine seems to be using a lot fewer resources.

I had been a little reluctant to upgrade to this release as I had read
about some problems.  But I think I would rather deal with those problems
than this one! :)

Thanks for the explanation!

 - Steve