Oh, I see! And I just saw your reply appear. 🙂
Zero Down Time
So I was wondering if anyone knew how to setup NodeBB to have zero down time. When something goes wrong, and NodeBB gets rebooted with Forever, there is a window of time that the site is down. Are there steps to having 2 NodeBBs running at the same time sharing the same database? (Database redundancy is a separate project for later.)
Are there steps to having 2 NodeBBs running at the same time sharing the same database? (Database redundancy is a separate project for later.)
Oh, now that's a fun idea... one could play around with pm2's built-in cluster functionality to run
xNodeBB instances, where
xis the number of CPU cores...
That would be so useful. Not to mention my downtime is horrible, lol this is partly due to the thousands of changes it goes through per week.
Having something like this would be critical to maintaining redundancy and improving overall downtime. Those 1-2 (sometimes 5~) min of downtime add up per week/month. I'm sure for larger sites this would serve to be extremely useful.
It would definitely be cool in terms of a development/production setup. Make your changes on the development version of the site, then push those changes to the live version. That would definitely minimize downtime.
I've only got 1 core at the minute, but still.
Yes one of the points of having 2 NodeBBs running would be for rolling updates of plugins that cause a reboot. I assume this would be very tricky or impossible to have rolling updates when a NodeBB Update requires schema changes to the Database.
I suppose it would be very much like downtime you'd see if you were upgrading any large CMS.
Have you tried using nginx to automatically send users to a pre-determined, statically-hosted error page if the connection to NodeBB 502s?