Hei folks,
I’d like to setup a cluster (two nodes running discourse) for high availibitly and load balancing.
How do setup such a cluster?
Thanks,
Jan
A user, yanosz, is seeking guidance on setting up a two-node cluster for high availability and load balancing. codinghorror pointed to the advanced install docs, which suggest deploying multiple containers in the web role talking to a database container.
However, yanosz had questions regarding database synchronization, suggesting Message Queuing or Postgres-Clustering as possible solutions. yanosz also inquired about the use of Docker and its interaction with the host system, as well as patching requirements.
Another user, zh99998, asked whether to use a shared Redis for all web nodes or separate instances. mpalmer advised using a central Redis, as it acts as a message broker for MessageBus traffic and having multiple instances would cause issues.
Hei folks,
I’d like to setup a cluster (two nodes running discourse) for high availibitly and load balancing.
How do setup such a cluster?
Thanks,
Jan
See the install docs at discourse/docs/INSTALL.md at main · discourse/discourse · GitHub which cover the advanced install. Basically you deploy multiple containers in the web role talking to a database container.
Hello,
thanks for your feedback. I read the manual before, but some thinks are not clear to me, yet :-/
Thanks, yanosz
Should I setup one shared redis for all web nodes, or setup multiply redis for each node separately ?
You need one central Redis. Redis isn’t (just) used as a cache for Discourse; it also acts as a message broker for all the MessageBus traffic, so if you had one Redis per node, things would get… weird. And broken.