|
Hello,
I'm running a two node OpenVPN cluster (let's call them node A and B)
using pacemaker and ldirectord.
For maintenance reasons I sometimes have to reboot a node.
Let's say I want to reboot node A.
To make this as convenient to the user as possible, I set the weight of
node A to 0.
Therefore all new connections go to node B and after some time there are
no users left on node A and I can reboot it.
But I have a problem, if node A's weight is set to 0 for maintenance
reasons and then node B goes down.
There are no destinations left and the whole cluster is down.
What I want: If node B goes down, let node A takeover regardless of the
weight.
My current solution:
I patched ldirectord making it run fallbackcommand when a node goes down
and there are no other destinations available. [1]
My fallbackcommand simply resets the weights.
Is there a better solution I'm not seeing?
Of course I would be happier with a vanilla ldirectord because this
would mean less maintenance burden.
Maintenance mode (maintenancedir setting) of ldirectord seems to be no
solution for me, because it kills the existing connections.
Thanks in adance,
Enno
[1]
https://github.com/egroeper/resource-agents/commit/d5e56f11df4320bbc19127ef56ab52d36f992c07
signature.asc
Description: OpenPGP digital signature
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users
| 