include-slaves - do not move slaves into write hostgroup even if the whole cluster went down.
General
Escalation
General
Escalation
Description
--include-slaves=host_name:port Add specified slave node(s) to ProxySQL, these nodes will go into the reader hostgroup and will only be put into
the writer hostgroup if all cluster nodes are down.
Slaves must be read only. Can accept a comma delimited list.
If this is used make sure 'read_only=1' is in the slave's
Even if the whole cluster went down do not move the async slaves into write hostgroup because that can be dangerous.
Issue The default behavior (if async slaves are used), is to add an async slave to the WRITE hostgroup if all PXC nodes are down.
Solution Make this into an option (--use-slave-as-writer=<yes/no>). This option is now required to be specified if async-slaves are specified when using proxysql-admin.
KennT September 13, 2018 at 10:20 AM
Edited
Totally agree, but we can't change existing behavior (if someone is depending on this). We will have to add an option for this. Maybe "-use-slaves-as-writers=<yes or no>" and add this as a required option if "-include-slaves" is used.
--include-slaves=host_name:port Add specified slave node(s) to ProxySQL, these nodes will go into the reader hostgroup and will only be put into the writer hostgroup if all cluster nodes are down. Slaves must be read only. Can accept a comma delimited list. If this is used make sure 'read_only=1' is in the slave'sEven if the whole cluster went down do not move the async slaves into write hostgroup because that can be dangerous.