Re: problem with UDP and conncect=negogiate

I just wanted to let anyone who cares know that I was able to remedy 
this issue with an upgrade:

Here is what I am now running:
ldirectord-2.0.6-2.fc5
pils-2.0.6-2.fc5
heartbeat-2.0.6-2.fc5
stonith-2.0.6-2.fc5

installed from yum.

the dns/udp/negotiate is working perfectly now. Thanks to Andersson 
Mattias  for the assistance.

-Jeff

Andersson Mattias wrote:
> Yes, I would say so.
>
> When it comes to DNS I would mostly prefer to load share such a service
> with DNS, i.e. multiple DNS-records. Notice that the load does not get
> balanced, but shared and that most well behaving client ask the next
> record when their first try fails. Unfortunate this is not true for
> exactly all clients however. 
>
>
> Back to ldirectord.
> Yes, as far as I can understand, have not tried it myself. The dns-check
> should have been implemented and working, just as you specified it,
> since version 1.80 of ldirector (24 feb 2004)...
> If you do not get it to work perhaps you could go with just a ping-check
> instead?
>
> /Mattias
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of Jeffery P.
> Humes
> Sent: Tuesday, July 18, 2006 8:59 PM
> To: [email protected]
> Subject: Re: problem with UDP and conncect=negogiate
>
> I take it back,  It does work with "checktype=on"  but this is defeating
>
> the purpose of my redundant realservers, is it not?
>
> If I bring one of the realservers down, it still has it marked at up in 
> the loadbalancer.
>
> -JPH
>
> Jeffery P. Humes wrote:
>   
>> Mattias,
>>
>> Yes, I had misspelled that option. thank you for catching that.
>> That did get rid of the error, but seemingly the checks for UPD/DNS to
>>     
>
>   
>> the realservers are not happening correctly
>> The VIP is setup correctly, however the servers are still being marked
>>     
>
>   
>> as "Quiescent".
>>
>>
>> ==> /var/log/ha-debug <==
>> heartbeat: 2006/07/18_13:40:09 debug: /etc/ha.d/resource.d/ldirectord 
>> ldirectord_dnsservers.cf start done. RC=0
>>
>> ==> /var/log/ha-log <==
>> heartbeat: 2006/07/18_13:40:09 info: local HA resource acquisition 
>> completed (standby).
>> heartbeat: 2006/07/18_13:40:09 info: Standby resource acquisition done
>>     
>
>   
>> [foreign].
>> heartbeat: 2006/07/18_13:40:09 info: Initial resource acquisition 
>> complete (auto_failback)
>> heartbeat: 2006/07/18_13:40:09 ERROR: Both machines own our resources!
>> heartbeat: 2006/07/18_13:40:09 info: remote resource transition 
>> completed.
>>
>> ==> /var/log/ldirectord_dnsservers.log <==
>> [Tue Jul 18 13:40:09 2006|ldirectord_dnsservers.cf] Starting Linux 
>> Director v1.77.2.11 as daemon
>> [Tue Jul 18 13:40:09 2006|ldirectord_dnsservers.cf] Added virtual 
>> server: 209.189.236.68:53
>> [Tue Jul 18 13:40:09 2006|ldirectord_dnsservers.cf] Added fallback 
>> server: 127.0.0.1:53 ( x 209.189.236.68:53) (Weight set to 1)
>> [Tue Jul 18 13:40:09 2006|ldirectord_dnsservers.cf] Quiescent real 
>> server: 10.1.2.207:53 mapped from 10.1.2.207:53 ( x 209.189.236.68:53)
>>     
>
>   
>> (Weight set to 0)
>> [Tue Jul 18 13:40:09 2006|ldirectord_dnsservers.cf] Quiescent real 
>> server: 10.1.2.206:53 mapped from 10.1.2.206:53 ( x 209.189.236.68:53)
>>     
>
>   
>> (Weight set to 0)
>>
>>
>>
>>
>> I have read that with "protocol=udp" if you do not specify the 
>> checktype, that it just marks the realservers as up all the time.
>> I have not gotten it to put them with any other weight than "0" for 
>> any checktype.
>>
>>
>>
>> -JPH
>>
>> Mattias Andersson wrote:
>>     
>>> Hi Jeffery, this one is probably really simple.
>>>
>>> I think you mean to say:
>>> receive="64.208.96.146"
>>> not
>>> recieve="64.208.96.146"
>>> Better be careful with the spelling when it comes to config-files, 
>>> they are not as forgiving as humans
>>>
>>> Hope that helps ;)
>>>
>>> /Mattias
>>>
>>> Jeffery P. Humes wrote:
>>>       
>>>> I have an UM3 setup running for quite some time now with no 
>>>> problems.  I love it.
>>>>
>>>> I have recently decided that it is a good Idea to loadbalance my DNS
>>>>         
>
>   
>>>> servers.  I have two BIND servers working correctly.
>>>> I have setup ldirecotor as follows:
>>>>
>>>> virtual=209.189.236.68:53
>>>>       real=10.1.2.206:53 masq
>>>>       real=10.1.2.207:53 masq
>>>>       fallback=127.0.0.1:53
>>>>       request="www.knworks.com"
>>>>       recieve="64.208.96.146"
>>>>       service=dns
>>>>       scheduler=wlc
>>>>       protocol=udp
>>>>
>>>> with the above configuration, I get errors in the logs and the load 
>>>> balancing for dns is not setup at all.
>>>> heartbeat: 2006/07/18_11:30:34 debug: Starting 
>>>> /etc/ha.d/resource.d/ldirectord ldirectord_dnsservers.cf stop
>>>> Error [] reading file /etc/ha.d/ldirectord_dnsservers.cf at line 6: 
>>>> Unknown command        recieve="64.208.96.146"
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> If I comment out the recieve="64.208.96.146" line from the 
>>>> configuration file, the loadbalancing is setup, however the 
>>>> realservers are marked as Weight "0"
>>>>
>>>> UDP  209.189.236.68:53 wlc
>>>>  -> 10.1.2.206:53                Masq    0      0          0        
>>>>  -> 10.1.2.207:53                Masq    0      0          0        
>>>>  -> 127.0.0.1:53                 Local   1      0          0     
>>>>
>>>> How do I get the UDP/DNS to work with ldirectord?
>>>>
>>>> --JPH
>>>>
>>>>
>>>>         
>>>       
>>     
>
>
>