[OpenSIPS-Users] A clustered mid_registrar instance increments ContactID when recovering data from seed node - bug or intention?

[Pavel Sindelka]

Hello users,

an active/standby cluster with a single common virtual IP address.

If a CPE registers for the first time (i.e. the AOR and its associated 
Contact have to be created), the ContactID is identical on both the 
active instance and the standby one.

If, however, one of the instances restarts and recovers the usrloc data 
from the other one via the cluster, the ContactID in the recovered 
contact becomes one higher than the original one, and future 
re-registrations of the CPE do not re-align it, although the expiration 
time does get updated on the standby instance. As a consequence, if the 
opensips instance that has restarted is/becomes the active one, the 
incoming calls to the CPE fail if mid_registrar's mode is set to 0 or 1, 
because the ctid parameter of the ruri does not match the ContactID.

Is anyone aware of some clever concept behind this behavior or I should 
stop thinking and just file a bug?

Thank you

Pavel