Questions on DCS availability

[Tiago-Anastacio]

Hello,

from patroni 3.1.2 release new parameter failsafe_mode was added
https://patroni.readthedocs.io/en/latest/dcs_failsafe_mode.html

Previously when the whole DCS (e.g. all etcdv3 nodes) was not reachable PostgreSQL primary was demoted and no other nodes could promote a postgreSQL standby until DCS comeback.

Now if the whole patroni's nodes are ok PostgresQL primary keeps reachable.

I have three questions ;

• what happenned with vip manager when the whole dcs (e.g. all etcd nodes) is not reachable, does it demote vip (on the node it was set) ?
• about configuration: dcs-endpoints: is it safer to specify a hostname instead of an IP ?
• Finally: is vip-manager compatible with a linux watchdog (I am thinking in case vip-manager service has been killed and at the same time vip is set on same node and should be removed )

Thank you