I've noticed that when the link to secondary DB is down, the primary DB spawns a high number of dataguard slave processes (RSM0 and NSV0), this was clear from the alert log.
When querying primary:
select count(*) from v$process;
gives a very large number, and sometimes it hits the maximum number of processes specified in the initialization parameter called "processes".
I can only guess that these processes are trying to connect to the standby database, but when I checked the parameter LOG_ARCHIVE_DEST_STATE_2 is set to DEFER.
How can I disable such behavior 'till the link gets active again?