In HA environment, where next hop monitoring feature is also configured, switch-over fails when the conditions represented with the following tests are met:
Environment: two different external linux machines, Linux1 (monitored by SSB1) and Linux2 (monitored by SSB2), respectively. SSB1 is the master.
1) The IP interface on Linux1 is turned off, and hence SSB1 could not ping it.
2) As a result, SSB1 soon becomes slave and SSB2 is activated as master.
3) Then the IP interface of Linux1 is turned on and this machine is available again.
4) Another switch-over is attempted (to bring up SSB1 as master) by turning off the IP interface of Linux2, but the awaited switch-over does not come.
Observations: before step 1) SSB2 was pinging Linux2 from its "ha0" (that is the redundant HA) interface, hence the linux machine was visible for the SSB. However, after step 3) the passive node SSB1 was pinging Linux1 from its unavailable eth0 interface.
The slave node should be able to ping the monitored linux, but in this case it does not happen.