[4539] in Athena Bugs
6.4R: /etc/rc doesn't work on Hesiod servers
daemon@ATHENA.MIT.EDU (Richard Basch)
Fri Mar 16 15:03:10 1990
Date: Fri, 16 Mar 90 15:02:49 -0500
To: bugs@ATHENA.MIT.EDU
From: Richard Basch <probe@MIT.EDU>
The database is too big, thus causing named to die because of the
default process limits.
Fix:
1) Add a BIGNAMED variable to rc.conf. If this value is not "false"
then it should invoke named as:
(/bin/csh -fc "unlimit;/etc/named /etc/named.boot") &
as opposed to the traditional method.
Reason: After a power hit, the Hesiod servers have to be restarted
by hand because the named grows to beyond 6Mb.
To further increase functionality, if the value of this variable is
not "false" and not "true", and is of the form xx.xx.xx.xx, it
should be taken as the host from which the server should get its
name-service. During the startup period, if this is not done, the
machine is "useless" until the named finishes its loading of the
database.
I realize that the variable name may not be what everyone would
consider correct, so I am open to better suggestions. Keep in
mind that there are very few cases of nameservice that require
this much space, and I don't think HESIODSRV is quite right...
(I am sure you probably recognize the text from a previous message, but
<jik> didn't like the first message, so I just cut out the relevant
section and am posting it again).
Note: this is being presented as a BUG, not a suggestion.
-Richard
