named.reload on bind 8.2 kills named

[Ronald Procopio]

Mark_Andrews at isc.org wrote:
> 
> > > Recently, doing a named.reload has been killing named on our primary
> > > NS. I am also getting these messages on my console when I reload:
> > > "named[16208]: ctl_server: bind: Address already in use". Those
> > > messages about the socket are new, I beleive. Bind is version 8.2
> > > running on a BSDI box that has 256 megs of Ram. I killed named then
> > > restarted it with the -d 3 flag and looked through the named.run file,
> > > and I did not see anything abnormal, but then, I might not know what
> > > to look for. A grep for fclose on named.run showed that it succeeded
> > > in loading every zone, at least.
> > > I frankly am new to this, and would appreciate any helpful
> > > suggestions, such how to debug this problem or any other steps I might
> > > take.
> > > --
> > > Pancho Cole, webmaster and hostmaster for AcadiaNet
> > >  http://www.acadia.net/     http://home.acadia.net/
> > >   Work (207)664-3100         webmaster at acadia.net
> >
> > On Solaris systems, this symptom means that the /etc/ndc "FIFO" file
> > must be removed before re-starting 'named'.  This bug has been fixed in
> > later versions of BIND.
> >
>         A Joseph says, upgrade to 8.2.1.  If after you have upgraded you
>         stil have this being reported you have two version of named
>         running.  BIND 8.2 reports this whenever the server was not
>         shutdown cleanly the last time.
> 
>         Mark
> > --
> > Joe Yao                               jsdy at cospo.osis.gov - Joseph S. D. Yao
> > COSPO/OSIS Computer Support                                   EMT-B
> > -----------------------------------------------------------------------
> > This message is not an official statement of COSPO policies.
> >
> --
> Mark Andrews, Internet Software Consortium
> 1 Seymour St., Dundas Valley, NSW 2117, Australia
> PHONE: +61 2 9871 4742                 INTERNET: marka at isc.org

Actually I have seen this with BIND 8.2.1 it seems to be because of
something that hasn't closed- I see this using the new ndc control
rpogram and doing a restart (AIX 4.3)  It seems to happen when a number
of transfers are taking palce at the same time of the restart (the
transfers aren't killed).  I corrected this be not using the restart
option and using stop and the start a minute later.




-- 
Bubble Memory, n.:
	A derogatory term, usually referring to a person's
intelligence.  See also "vacuum tube".
=================
Ronald Mark Procopio		procopio at ims.advantis.com