Latest CVS on solaris 2.6 x86

Jeremy Allison jeremy at valinux.com
Tue Feb 27 16:05:07 GMT 2001


Joe Meslovich wrote:

>         I have not been able to join a W2K workstation to that domain
> since the restructuring of the samr code. I have even tried builds from
> the 24th and I am unable to join. The code from the 24th is all around the
> most stable with the exception of being unable to join the domain. All of
> by builds since the 24th have been rather unstable. I initially
> contributed this lack in stability to the fact that Jeremy was redoing the
> RPC code and splitting it into implementation and interface modules, but
> he supposedly finished that update sometime last night my time. Anyway
> with the code as of this morning I still cannot join my samba domain,

I did test this a couple of days ago after JF and I finished updating
the samr code. I will test it again.

Remember, living out of CVS is *dangerous*. It should always compile,
but it might not always work. That's why I release the tarball snapshots.

> and
> when I try to do so the smbd process on the server steals up to half of my
> processor cycles ( old dual processor P166 ) and takes up 207 MB of
> resident memory. After much processing the smbd process returns an error
> to the W2K wks saying that he procedure is out of range. I do not have
> debugging turned on, but shall do shortly to produce a log.
> 
>         I know that Jeremy warned that there would be memory leaks, but
> the CVS builds were more or less stable up until the start of the RPC
> restructuring on the 24th.

Well I know it's not leaking memory, as I'm checking this
with insure. It may be however not freeing memory (different
problem) - this would happen if the talloc was context was
not being cleaned up correctly at the end of an rpc call.

I'll check this.

Jeremy.

-- 
--------------------------------------------------------
Buying an operating system without source is like buying
a self-assembly Space Shuttle with no instructions.
--------------------------------------------------------




More information about the samba-technical mailing list