Heartbeat gets paused, but doesn't resume anymore

["Martijn Vogel" <lvs-users@xxxxxxxxxxxxx>]

Hi all,

While I'm mostly a reader of this mailinglist and learn a lot from problems 
(and offcourse solutions) others have, I now seem to have problem noone has had 
before. Please prove me when wrong, I might of have missed the answer somewhere.

I'm using heartbeat 1.99.2 ( http://www.linux-ha.org/download/ ) in a dual 
loadbalancer setup for LVS-NAT (active/standby). After some succesful takeovers 
heartbeat falls into a state of pausing because of congestion (a feature added 
recently if I check the code listing). Unfortunately it never resumes again :(

The line in /var/log/ha-log which explains it all:

heartbeat[22877]: 2005/03/08_14:23:11 info: all clients are now paused

The heartbeat is on a serial cable at 19200 baud. The problem also occurs when 
doing this with mcast or with mcast + serial cable.

- What makes the congestion occur so all clients are paused?
- Why don't they resume operations anymore?

Thanks for you help!

Martijn