Re: single point of failure in a highly available cluster of web servers.

On Tuesday, Mar 18, 2003, at 16:44 US/Eastern, Rama K. McIntosh wrote:
> I am trying to eliminate a single point of failure outlined below (I 
> am new to Wackamole so please forgive me if I am missing the basics).
>
> My concern is Wackamole seems to be aware of computer servers crashing 
> but not the web server crashing.  Is there anyway to have Wackamole 
> take over the IP of a server if httpd crashes, not the machine itself.
>
> For example I have two backhanded servers with Wackamole to service a 
> virtual IP.  If machine A has the IP and it crashes, the Wackamole 
> will have Machine B take over.   However, if httpd on Machine A 
> crashes while Machine A is still running fine, my web farm is now dead 
> even though Machine B is functioning fine.
>
> Is there any way to eliminate this single point of failure?

There is a program distributed with wackamole called wacktrl.  It 
allows you to fail a machine administratively.

So, right a service tester (in perl or whatever) and if the service 
tester comes back with a negative on machine A, have it run wackatrl -f 
and when it starts working again have it run wackatrl -s.

-- 
Theo Schlossnagle
Principal Consultant
OmniTI Computer Consulting, Inc. -- http://www.omniti.com/
Phone:  +1 410 872 4910 x201     Fax:  +1 410 872 4911
1024D/82844984/95FD 30F1 489E 4613 F22E  491A 7E88 364C 8284 4984
2047R/33131B65/71 F7 95 64 49 76 5D BA  3D 90 B9 9F BE 27 24 E7