Problems with 'ndc restart' on BIND 8.3.3

[Ian Bedson]

Hi Everyone,

I have compiled and installed BIND 8.3.3 on my servers in response to the
recent security alerts. However I am now seeing a new behaviour from the ndc
command due (I think) to the way that ndc has been written in this release.

In the past I have always been able to use the 'ndc restart' command
whenever I have updated my primary servers. Now I find that 'ndc restart'
does not work in the same way any more. So I have had to modify my scripts
to execute an 'ndc stop' followed by an 'ndc start'.

The problem is that because I am running these commands from inside shell
scripts that are executed from cron, I am using the exit code from ndc to
detect if there were any errors. With BIND 8.3.3 when I use ndc to restart
named it returns an exit code of 1 and writes the message, "ndc: error: name
server has not restarted (yet?)" to the console.

I guess that these changes have been made so that on servers with a large
number of zones you can execute the 'ndc restart' command and get the
command prompt back again quite quickly, while the job of restarting named
proceeds silently in the background. I have verified that if I wait long
enough, the server will restart successfully.

So while this message isn't really a request for help (I am quite content to
stick with my 'ndc stop' / 'ndc start' strategy) I just want to verify that
the new behaviour that I am seeing is correct. Otherwise it will mean that
my build of the new code is broken.

By the way, the code was compiled using GCC 2.95.2 running on servers
running HP-UX versions B.10.20 and B.11.00

Regards,

Ian