Keepalived takes several minutes to recover in a particular situation

Posted by NathanE on Server Fault See other posts from Server Fault or by NathanE
Published on 2011-02-28T15:10:46Z Indexed on 2011/02/28 15:26 UTC
Read the original article Hit count: 314

Filed under:

keepalived

I've setup Keepalived for a master-slave style virtual IP and it seems to work well.

Both are hosted in almost identical VMs.

If I "pause" the VM that is running the Master. The Slave will take over, as expected, almost instantly.

However if I then "unpause" the VM that runs the Master. The virtual IP will stop responding the pings. And it takes a good 4 or 5 minutes for it to start pinging again.

It seems to be getting desynchronised due to the nature of the way I'm testing it (by pausing/unpausing the VMs).

I admit that pausing and unpausing VMs is a slightly dodgy way to test this. But it has raised a concern for me that there could be other scenarios that cause the same undesirable behaviour.

Is this expected / by design?
Is there anything I can do to the config to improve it?

Thanks.

Related posts about keepalived

keepalived questions (requirements, abilities, limitations)

as seen on Server Fault - Search for 'Server Fault'
1) What are keepalived's (physical/network) requirements? Does the two (or more) keepalived' nodes need to be connected to the same switch? (something related to broadcasting maybe). 2) Can keepalived nodes run on different networks, "internet" networks? 3) Is keepalived depend on the router? (as… >>> More
keepalived issues on xen domU

as seen on Server Fault - Search for 'Server Fault'
Hi, I cannot manage to run keepalived correctly on xen domU. I am following this link for configuration, and it works great on some local VM (running with KVM). If I set up the exact same configuration, but on xen domU, it does not work: both servers do not see each other and decide to be master… >>> More
Keepalived takes several minutes to recover in a particular situation

as seen on Server Fault - Search for 'Server Fault'
I've setup Keepalived for a master-slave style virtual IP and it seems to work well. Both are hosted in almost identical VMs. If I "pause" the VM that is running the Master. The Slave will take over, as expected, almost instantly. However if I then "unpause" the VM that runs the Master. The virtual… >>> More
How to configuration keepalived on Amazon EC2?

as seen on Server Fault - Search for 'Server Fault'
I rad some article. Keepalived over GRE tunnel for failover on VPS environment http://blog.killtheradio.net/how-tos/keepalived-haproxy-and-failover-on-the-cloud-or-any-vps-without-multicast/ but, I don't know how to configuration? and How to call this architecture? only I Know that How to config… >>> More
Both servers running keepalived become master

as seen on Server Fault - Search for 'Server Fault'
After a network failure,both servers running keepalived become master. When the network is reestablished, both keep the MASTER state. What could be causing it? Edited: Another information that might be relevant, each server has two NICs. Here is the virtual instance configuration: vrrp_instance… >>> More

Developer IT