Auto-healing for unavailable servers
Using gauges, set up auto-healing in a container deployment to address an unavailable server.
Steps
-
Configure one or more of the gauges described in Server availability.
-
Configure the gauges to trigger the UNAVAILABLE status.
By default, the gauges do not trigger the UNAVAILABLE status.
As discussed in Endpoint Average Response Time (Milliseconds) gauge and HTTP Processing (Percent) gauge, use the
dsconfig
command to adjust the following values for your environment. Each system is different so you might need to adjust the values several times to determine your ideal configuration.-
For the
Endpoint Average Response Time (Milliseconds)
gauge, setcritical-value
. -
For the
HTTP Processing (Percent)
gauge, set bothcritical-value
andserver-unavailable-severity-level
.
-
-
Configure the container orchestrator to use the
available-or-degraded-state
endpoint to detect whether the server is alive.For information about the endpoint, see paz_server_status.adoc#section_kgv_n53_zkb.