Login | Register For Free | Help
Search for: (Advanced)

Mailing List Archive: mod_backhand: Wackamole

wackamole convergence speed

 

 

mod_backhand wackamole RSS feed   Index | Next | Previous | View Threaded


john.jolet at fxfn

Apr 7, 2003, 2:31 PM

Post #1 of 4 (1075 views)
Permalink
wackamole convergence speed

I've got wackamole working great for ip failover. Unfortunately, it's
working TOO great. apparently, we don't want the address to fail over any
quicker than one minute. apparently, we've got some networking issues that
we hope to sweep under the rug by not failing over at the first sign of
trouble. is there a configuration parameter that will cause it to not fail
the interface over unless it's been a minute?


caudy at jhu

Apr 7, 2003, 3:22 PM

Post #2 of 4 (1014 views)
Permalink
wackamole convergence speed [In reply to]

There's no solution for that built into wackamole. Previous attempts to
alter the failure detection time have all been in the other direction,
and have involved modifications to timeouts within the Spread toolkit.
Modifications in the opposite direction shouldn't be impossible, and may
suit your application here. On the other hand, it would probably be
easier to just add a delay that works similar to the balance timeout
between the receipt of a regular membership message from Spread and the
beginning of the GATHER state.

--Ryan Caudy

John Jolet wrote:

>I've got wackamole working great for ip failover. Unfortunately, it's
>working TOO great. apparently, we don't want the address to fail over any
>quicker than one minute. apparently, we've got some networking issues that
>we hope to sweep under the rug by not failing over at the first sign of
>trouble. is there a configuration parameter that will cause it to not fail
>the interface over unless it's been a minute?
>
>_______________________________________________
>wackamole-users mailing list
>wackamole-users [at] lists
>http://lists.backhand.org/mailman/listinfo/wackamole-users
>
>
>


john.jolet at fxfn

Apr 7, 2003, 3:25 PM

Post #3 of 4 (1006 views)
Permalink
wackamole convergence speed [In reply to]

in other words modify the code? Can do, but would prefer to avoid it, if
there's a config option. Can anyone give me a rundown of what the maturity
rate and balance interval in the config really mean in practical terms?

-----Original Message-----
From: Ryan Caudy [mailto:caudy [at] jhu]
Sent: Monday, April 07, 2003 5:23 PM
To: wackamole-users [at] lists
Subject: Re: [Wackamole-users] wackamole convergence speed


There's no solution for that built into wackamole. Previous attempts to
alter the failure detection time have all been in the other direction,
and have involved modifications to timeouts within the Spread toolkit.
Modifications in the opposite direction shouldn't be impossible, and may
suit your application here. On the other hand, it would probably be
easier to just add a delay that works similar to the balance timeout
between the receipt of a regular membership message from Spread and the
beginning of the GATHER state.

--Ryan Caudy

John Jolet wrote:

>I've got wackamole working great for ip failover. Unfortunately, it's
>working TOO great. apparently, we don't want the address to fail over any
>quicker than one minute. apparently, we've got some networking issues that
>we hope to sweep under the rug by not failing over at the first sign of
>trouble. is there a configuration parameter that will cause it to not fail
>the interface over unless it's been a minute?
>
>_______________________________________________
>wackamole-users mailing list
>wackamole-users [at] lists
>http://lists.backhand.org/mailman/listinfo/wackamole-users
>
>
>



_______________________________________________
wackamole-users mailing list
wackamole-users [at] lists
http://lists.backhand.org/mailman/listinfo/wackamole-users


caudy at jhu

Apr 7, 2003, 3:41 PM

Post #4 of 4 (1015 views)
Permalink
wackamole convergence speed [In reply to]

Maturity is a timeout for when Wackamole daemon's run initially. They
don't grab any IPs the first time until the timeout occurs, or until
they hear from someone who is already mature. Balance is just the
length of time between rounds of balancing IP addresses between daemons
(if you configure it to do so incrementally). Neither helps in your
situation, as far as I can imagine. :-(

--Ryan

John Jolet wrote:

>in other words modify the code? Can do, but would prefer to avoid it, if
>there's a config option. Can anyone give me a rundown of what the maturity
>rate and balance interval in the config really mean in practical terms?
>
>-----Original Message-----
>From: Ryan Caudy [mailto:caudy [at] jhu]
>Sent: Monday, April 07, 2003 5:23 PM
>To: wackamole-users [at] lists
>Subject: Re: [Wackamole-users] wackamole convergence speed
>
>
>There's no solution for that built into wackamole. Previous attempts to
>alter the failure detection time have all been in the other direction,
>and have involved modifications to timeouts within the Spread toolkit.
>Modifications in the opposite direction shouldn't be impossible, and may
>suit your application here. On the other hand, it would probably be
>easier to just add a delay that works similar to the balance timeout
>between the receipt of a regular membership message from Spread and the
>beginning of the GATHER state.
>
>--Ryan Caudy
>
>John Jolet wrote:
>
>
>
>>I've got wackamole working great for ip failover. Unfortunately, it's
>>working TOO great. apparently, we don't want the address to fail over any
>>quicker than one minute. apparently, we've got some networking issues that
>>we hope to sweep under the rug by not failing over at the first sign of
>>trouble. is there a configuration parameter that will cause it to not fail
>>the interface over unless it's been a minute?
>>
>>_______________________________________________
>>wackamole-users mailing list
>>wackamole-users [at] lists
>>http://lists.backhand.org/mailman/listinfo/wackamole-users
>>
>>
>>
>>
>>
>
>
>
>_______________________________________________
>wackamole-users mailing list
>wackamole-users [at] lists
>http://lists.backhand.org/mailman/listinfo/wackamole-users
>
>_______________________________________________
>wackamole-users mailing list
>wackamole-users [at] lists
>http://lists.backhand.org/mailman/listinfo/wackamole-users
>
>
>

mod_backhand wackamole RSS feed   Index | Next | Previous | View Threaded
 
 


Interested in having your list archived? Contact Gossamer Threads
 
  Web Applications & Managed Hosting Powered by Gossamer Threads Inc.