tomcat-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From fredk2 <fre...@gmail.com>
Subject Re: Apache fails to start when .host does not resolve.
Date Tue, 08 Apr 2008 21:41:08 GMT

Hi
Thank you for your reply - as alway I appreciate. 
I can understand your point very well and I did dismiss the original
complaint as such - a one-off, a minor issue.
There is one silly anecdote: assume that you have a heavily shared Apache
with a bunch of workers... one day one of the friendly *sharee* decides that
he does not need one of his hosts anymore and forgets to tell the server
admin. A later restart affects all the services hosted because of this
...how to say it... oversight:-).  To get apache up again you probably need
to logging physically to the server and vi the config file (some pain).

Rgds
Fred


Rainer Jung-3 wrote:
> 
> fredk2 schrieb:
>> Hi,
>> 
>> when you set a load balancer (mod_jk v1.2.26) with 2 workers:
>>     worker.myWorker.type=lb
>>     worker.myWorker.balance_workers=tc1Worker, tc2Worker
>> 
>> and one of the worker's host cannot be resolved:
>>     worker.tc2Worker.host=mytest.mydom.com
>> 
>> Then Apache will not start.
>> 
>> - since the other worker is 'good', should'nt we let apache start with a
>> nice error message informing you that one worker could not resolved (done
>> today) and jk will disable the worker and let apache start anyway?
>> 
>> - Adding a disable does not fix the issue it seems.  Only removing the
>> worker from the list does.
>>     worker.tc2Worker.activation=d
> 
> Yes that's the way it was implemented intentionally.
> 
>> The idea is that if DNS fails to resolve a host (some admin mistake) and
>> Apache was restarted automatically for unrelated reasons, it would fail
>> to
>> come back online without a manual intervention - editing a configuration
>> file (which might be a challenge).
> 
> In general I think it's better to not start, if a worker is completely 
> broken. Yes, in your case it would provide a problem with regular 
> restarts, but we could argue, that if your DNS is not rock solid, but 
> your production systems depend on IP address resolution, then you should 
> either use the IPs in the workers host attribute, or add the few systems 
> you need to contact from your httpd to your hosts file. Of course that 
> contradicts the major goals of DNS, but if you want to stick to those 
> you really need a rock solid DNS (which is not that hard). Admin 
> failures when managing critical DNS data are really major incidents.
> 
> If we allow unresolvable host names, it's very likely that users will 
> not notice the resulting error messages during startup. And I think it's 
> much more likely that people simply put typos into the host names than 
> the occurence of local DNS failures.
> 
> This seems to be grey area, where there's no obviously best solution.
> 
>> Rgds - Fred
> 
> Regards,
> 
> Rainer
> 
> ---------------------------------------------------------------------
> To start a new topic, e-mail: users@tomcat.apache.org
> To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
> For additional commands, e-mail: users-help@tomcat.apache.org
> 
> 
> 

-- 
View this message in context: http://www.nabble.com/Apache-fails-to-start-when-.host-does-not-resolve.-tp16422828p16574355.html
Sent from the Tomcat - User mailing list archive at Nabble.com.


---------------------------------------------------------------------
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: users-unsubscribe@tomcat.apache.org
For additional commands, e-mail: users-help@tomcat.apache.org


Mime
View raw message