Forge / DO servers fail for no visible reason

Hello,

we are running a lot of DO servers using the Forge. At the beginning of this year it started to happen to us that one of the servers stopped communicating without any obvious reason: error connection timeout in the browser, ssh connection timeout. After rebooting the server via DO everything returned to normal. We went through everything, every single log, load graphs - nothing suspicious anywhere. We have contacted DO support several times, the answer is mostly similar: We have checked the droplet, it is fine, the cause is usually your application, settings... blah blah.

We have web monitoring in place, we have created an automatic system to restart the server if it is unavailable. But for me, it's just addressing the effect and not the cause.

We've been using Forge/DO for many years, we didn't have these issues until this year. The applications we have on the servers where this problem occurred are diverse. From simple Laravel applications, custom PHP applications, WP websites. In short, we haven't come up with any rule of thumb to work from. The last time this happened was this Sunday on a server running an ordinary Laravel application. The server was minimally loaded, no recent updates of that app.

I would be glad for any tips on where and how to look for the cause.

Thanks.