fully scaled architecture with 6 CM servers and 6 publisher servers. On publisher only publisher and transport services are running.
Recently, we came across a strange scenario where
one publisher rendered the publish package. After some time we received the failure for the publishing transaction and examining the logs the we found out the
failure was for polling notification and strange part was it was logged in different publisher than the one that performed the render operation.
In which scenario, publisher switch / failover will happen? Any help in this matter is appreciated.