|
First, thanks to Joseph Mack for his advice. I've got a lot to work with
now.
Joseph Mack wrote:
> In general this can't be done - you loose the connection and have to
> start again. It's on the todo list.
Michael Sparks wrote:
> ? I must read that differently to you :-) ?
>
> current realservers : A,B,C
> servers to swap into the system instead D,E,F
>
> * Add servers D,E,F into the system all with fairly high weights (perhaps
> ramping the weights up slowly so as not to hit them too hard:-)
> * Change the weights of servers A,B,C to 0.
> * All new traffic should now go to D,E,F
> * When the number of connections through A,B,C reaches 0, remove them from
> the service. This can take time I know but...
>
> Has the required effect.
Yes, this is exactly what I want to do. Except that I'd like to do it such
that all the servers (well only two) are really running on a single
machine. So the director and real-server addresses are all on one box. The
idea behind this is that its efficient and allows "feathering down" a pool
of servers bound to a single port or address and bringing up a new pool
bound to a different port or address without clients noticing or modifying
any of the server code. Cycling servers every now and then is needed both
to prevent resource exhaustion and to change configurations. (Yes, better
written server software usually has some support for this internally. But
some does not...)
For HTTP, I can do this with a proxy server like squid. I'd like something
a little more general (and perhaps more efficient for cases where data
cannot be cached).
Outside of the above setup, I'm also looking at using LVS for
high-availability using multiple real-servers.
-Z-
----------------------------------------------------------------------
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
To unsubscribe, e-mail: lvs-users-unsubscribe@xxxxxxxxxxxxxxxxxxxxxx
For additional commands, e-mail: lvs-users-help@xxxxxxxxxxxxxxxxxxxxxx
|
