This patch adds an option to wait for replicas to come online (respond to ping) during the replication host sanity check phase.
This is relevant when a replicated cluster is brought offline completely, and then restarted. The first container to restart will fail, but subsequent containers will find its hostname routable. With this option turned on, instead of failing when the replica hostnames do not resolve, the initialization script will wait for them to become available before proceeding.
If the cluster configuration changes while the cluster is offline (e.g. one node is renamed or removed), the state in LDAP will need to fixed by hand in any case.
tiredofit
commented
4 years ago
This patch adds an option to wait for replicas to come online (respond to ping) during the replication host sanity check phase.
This is relevant when a replicated cluster is brought offline completely, and then restarted. The first container to restart will fail, but subsequent containers will find its hostname routable. With this option turned on, instead of failing when the replica hostnames do not resolve, the initialization script will wait for them to become available before proceeding.
If the cluster configuration changes while the cluster is offline (e.g. one node is renamed or removed), the state in LDAP will need to fixed by hand in any case.