[CentOS] Bind problem - rndc key (after update?)

Sat Nov 18 19:59:34 UTC 2006
kadafax <kadafax at gmail.com>

William L. Maltby wrote:
> On Sat, 2006-11-18 at 16:03 +0100, kadafax wrote:
>   
>> Hi list,
>>
>> here is what happened:
>> today I noticed some resolution's problems on my network. I did a 
>> "service named status"  and here was the output:
>> # /etc/init.d/named status
>> rndc: connection to remote host closed
>> This may indicate that the remote server is using an older version of
>> the command protocol, this host is not authorized to connect,
>> or the key is invalid.
>>
>> In the named's log, several entries like that:
>> general: error: invalid command from 127.0.0.1#42033: bad auth
>>
>> I am not using the key's authentication on my chrooted bind dns and it 
>> was working great so far.
>>
>> Searching on rndc's files in /etc I've found mismatch for the key value 
>> in /etc/rndc.conf and /etc/rndc.key. There was also a rndc.key.rpmnew file.
>> After giving the good value for the key entry (I've copied-pasted the 
>> value from the .key file), the bind daemon seems to be happy now.
>>
>> My question is how things get broken because I didn't touch the bind's 
>> config files for a year or so (only the zone files, sometime) ?
>>     
>
> Search the Centos archives for a complete explanation. Basically, a
> recent update changed the configurations (that's why you have an .rpmnew
> file) so that your rndc keys no longer match.
>   
the rpmnew file means that an update created a *new* file. A .rpmsave 
file means that an update had renamed and replaced the original file. 
You really should know that...
> After an update, it's always a good idea to updatedb and then locate
> *.rpmnew and/or *.rpmsave.
>   
No bind update since 4.4 fork. I move from 4.3 to 4.4 since some times, 
and the problem only appears now.
> The *potential* for the problem was reported *very* early after the 4.4
> (?) update and those who watch the lists regularly avoided problems.
>   
useless...
>   
>> <snip>
>>     
The .rpmnew file is independent of my problem in this case.
Is it possible that a process had changed the original rndc.key value? 
Thanks.