Discourse setup with redundancy

What is the recommendation to run discourse in a High Availability setup? What is the best practice?

Is anyone running Discourse in HA setup?

I imagine a lot of VPS servers are by their very nature, Highly Available, because they are running on a big ole HA back end server cluster. But worth checking with your potential service provider.

“Bare metal” servers may be different but you can surely set up HA proxies? Not tried this though. Defer to others more experienced here …

In any case, in the 1.5 years of running 2 Discourse sites on VPS’s, neither has gone down ONCE! (ok ok one of them has only been live half of that time)

(NB I have restarted services many many times due to umpteen upgrades so bear that in mind, but not because of failures)

(hats off to everyone involved in making that happen at every layer of that challenge, seriously! In my career I’ve used Windows Servers & Oracle Web Logic which were soooo unreliable and rebooted at least every few weeks and sometimes every fortnight!)

A few things to consider:

  • How critical is your content? Is a daily backup enough? (complete rebuild is dead easy if you have a forum backup and a copy of app.yml and would take an experienced guy 30 mins max). Can you afford to lose a day’s information? If so then the out of the box solution is perfect.
  • Downtime when upgrading is measured in seconds I believe … which is why @codinghorror laughed at me for asking how to send out an old fashioned ‘downtime’ email to all users. Old habits die hard.