Hazelcast misbehaving with Docker
https://github.com/bitsofinfo/hazelcast-docker-swarm-discovery-spi
Genesys (production) is now configured as two separate services (server and standby) that try to expose the correct names using service names. While this works most of the time, we have a cluster brain-split every few weeks, causing 403 HTTP errors as _crsf tokens are out of sync.
Maybe the best approach would be to add another Hazelcast instance (to 3 in total) as a docker service. This may prevent issues with merging after a brain split.