At approximately 20:10 we've detected an outage with one of our database clusters. This issue interrupted the internal replication procedures which in effect created a split-brain problems when it comes to various data I/O operations (SIP registrations, concurrent calls checks, etc).

The issue originated from our VoicePortal, which was temporary disabled in order to restore stability of the affected database cluster. During the recovery some SIP registrations were terminated in order to speed up the process.

Our voice services were fully restored at 20:16.

SIP registrations for all subscribers were restored between 20:16 and 20:21.

The VoicePortal was restored at 20:23.

Voice Portal recovered.

Voice Portal went down.