I’ve just setup two ATCs in HA by simply starting two with identical settings but the
--tsa-peer-ip and using a single DNS entry with two assigned IP addresses for the workers to use and an nginx as load-balancer for the web UI only.
The few short tests I’ve done so far seems to indicate that everything works fine. However, a colleague pointed me to the topology page (https://concourse-ci.org/topology.html#web-node) that says:
Highly available: yes;
webnodes can all be configured the same (aside from
--peer-url) and placed behind a load balancer.
Which would imply that a load balancer is mandatory. But the install page contradict this (https://concourse-ci.org/install.html#running-web-nodes):
webnode can be scaled up for high availability, and they’ll also roughly share their scheduling workloads, using the database to synchronize. This is done by just running more
webcommands on different machines, and optionally putting them behind a load balancer.
So, my questions are:
- Do I need to connect the workers to the ATCs through a load balancer or not?
- What’s the purpose of the
--tsa-peer-ip? It’s mentioned nowhere in the docs. Can I safely ignore it?