Discourse Crashing on DigitalOcean - twice in one week

(Bcguy) #1


I’m running the latest Discourse release on Digital Ocean on their 2 GB service.

The server has stopped working twice in the past week and I’m wondering what could be wrong. I’ve had to go in and do a soft restart whenever this happens.

This is the first time in 1 year of operation so it seems like maybe I’m bumping up against some sort of limit.

We have about 200,000 messages in 20,000 topics. We get about 30,000 api requests a day. I have it set up to do two backups and we’re really down on the backup space left:

Here is what I’m seeing in the Digital Ocean reports panel:

And here are the long term trends on my forum:

Also - is there a way to get automatic notifications from Digital Ocean when the forum software goes down? Right now I have to wait until I discover it when I happen to check the forums.

Any ideas? (I’m not a software engineer).

(Tobias Eigen) #2

I use pingdom to notify me when my site is down - it’s free: Free Website Monitoring | Pingdom


Strangely enough, this happened to us on our server for the first time ever a few days ago. We use Vultr though. I couldn’t SSH into the server at all, I had to force a shutdown using the admin page in Vultr.

(Dean Taylor) #4

It looks you are running out of space here, strange things will always happen when you run out or just low disk space wise.

You should really target having 2x the space of a single complete backup free as a minimum.

(Jeff Atwood) #5

I would definitely go up a level on your Digital Ocean plan – more disk space, etc. Do not forget to edit app.yml to give the database more memory.


If you have that much space currently allocated for backups, while at the same time have less than 200MB left, perhaps an easy way to gather if the issue is solely related to a lack of disc space would be downloading some of the backups to local storage, deleting them off the VPS, and checking the stats and feel of the site over the next few days. There is a loading of CPU recently in those stats, but I’ve noticed this can happen as a system tries to make due with resource constrictions elsewhere. I’ve seen it with memory restrictions, but don’t think it’s a stretch to imagine it can happen from disc space restrictions as well.

(Michael - DiscourseHosting.com) #7

How about your memory usage? Can you use SSH to get a shell, type free and paste the output?