network has gone down again, and I cannot figure out why

[David G. Miller]

Ric Moore <wayward4now at gmail.com> wrote:

>On Mon, 2006-09-25 at 09:34 -0700, Mike Wright wrote:
>  
>
>>> Ric Moore wrote:
>>    
>>
>>>> > I have a static IP, and with the DSL modem set to "pass through" mode,
>>>> > my linux "crate" is doing everything... and blowing up. Been working
>>>> > rock solid, went through the update, a report is made about something
>>>> > missing and I go into scattered chickens mode becauce weird stuff is
>>>> > happening, like the domainname being added as an alias to localhost
>>>> > minutes after I fix it.
>>>      
>>>
>>> 
>>> I did lookup on your hostname and got an IP but a reverse lookup on that 
>>> IP gives adsl-070-145-234-214.sip.int.bellsouth.net.  I wonder if the 
>>> problem arises because the forward and reverses conflict, and perhaps 
>>> one of the configuration programs is "fixing" it for you?
>>    
>>
>I believe I am OK now, If you would try that again to see what pops up,
>I'd appreciate it. I am back on my static IP, it's resolving quickly and
>seems to be at full speed. 
>
>> It's brute force but you could try "chattr" to make your hosts file 
>>> immutable.
>>    
>>
>I'm afraid to touch it, now that it's working. If it wasn't the bind
>packages, then using the FC6 setup* rpms must have done the trick. Ric
>
Just a WAG but it appears that one of the rpms in the latest batch of 
updates does something really strange with /etc/hosts.  I updated my FC5 
box yesterday and /etc/hosts now looks like:

[dave at spindle ~]# cat /etc/hosts
# Do not remove the following line, or various programs
# that require network functionality will fail.
192.168.255.254   mutilate.local.davenjudy.org    mutilate
127.0.0.1               spindle.local.davenjudy.org spindle 
localhost.localdomain localhost

The first line is my NIS server so luckily the update didn't mess with 
that.  I can't imagine any circumstances under which I would modify the 
127.0.0.1 entry to look like the above.  At least I didn't run into the 
problem of the bogus line being "sticky".  I still don't have an answer 
to that problem.

Also, looking at what logwatch reports, I noticed the following:

 Packages Erased:
    bind-config 

Another WAG but it appears to have been replaced by:

 Packages Installed:
    sane-backends-libs.i386 1.0.18-2.fc5
    kernel.i686 2.6.17-1.2187_FC5
    caching-nameserver.i386 30:9.3.2-33.fc5 <-- This
    libgconf-java.i386 2.12.4-2.fc5


Cheers,
Dave

-- 
Politics, n. Strife of interests masquerading as a contest of principles.
-- Ambrose Bierce