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 )
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 ;
Thank you