On 1/20/23 12:14, Matthew Castrigno wrote:
Thank you Shawn.
My configuration is standalone mode.
We are going to use HAproxy and need a metric to determine when to
switch between two servers. The severs are kept "in sync" with each
other by copying the cores from one to another. This works for our use
case which only indexes new content on a schedule which we control.
I believe it also returns an HTTP response code that's an error type if
the query fails. Because your ping handler has a q parameter, I think
you can reasonably rely on the OK status as a good check, and I *THINK*
you can also rely on the http response code.
I like to use the healthcheck option, and I do know that it returns a
503 response code if the healthcheck flag is not enabled. This is a
really good way to make it so that healthy servers are removed from
rotation in case you want to force queries to go elsewhere.
Thanks,
Shawn
