o, we've been running in the cloud now for a while, and after some initial problems, things seem to be calming down.
I'm still investigating issues of instability when the server is running for a long time. I'll keep you posted.
For your information, I'm running on a PaaS (Platform as a Service) framework called Jelastic (https://jelastic.com/).
| Old setup (VPS) | New setup (Cloud) | Comments |
|---|---|---|
| flexibility, I was free to do as I pleased | inflexible, node setup is done for you | |
| high maintenance | low maintenance | Related to previous point. |
| slow | fast | Most users are Americans, and the game is now housed in a Data Center in Atlanta of all places. |
| simple support, no statistics | extensive support, detailed statistics | |
| steady predictable cost | only pay for resources you use | |
| inefficient software, but who cares? | inefficient software is costly! | Related to previous point. |
| payment via bank account | attached to my PayPal | That's right! If you donate to my PayPal it goes straight to the server. |
| my server, mine, no touchy! | collaboration is possible/ownership is transferable | Nice to know if something were to happen to me. |
To be fair, my old setup did have several options, I just didn't know about them. At the time of starting my contract period, at least, they were not available.
Installing/replacing a node changes the IP address, which means I have to change the DNS records of karchan, which means people have to wait for the DNS change to trickle down to them.
I accidentally installed an old version of the application server. That needed to be corrected. My mistake.
I installed NGINX to be able to redirect HTTP traffic to HTTPS, which seemed fine, however, it also caused the Websocket to be closed after 1 minute of inactivity. So for now, I removed the NGINX again.
The domain used in the application server is not suited for production use, and needed to be changed. It might have been the cause for the server to stop responding. I increased the EJB Bean pool to 128 (up from 32), the http thread pool to 50 (up from 5) and turned off the http listener for now (until I can get the redirect fixed). Transaction timeout set from 0 (disabled) to 300 seconds just in case.
I plan on changing some more things about the settings in the application server, but these were the important ones.
Fixed the pictures of the rooms, a lot of them were prefixed with http://www.karchan.org and http://karchan.org and they should not be. As a friendly reminder, if, for instance, you need to reference the picture well.gif on the server itself, a relative url "/images/gif/well.gif" would be perfectly fine.
Removed a lot of old user methods containing the old xml way of working. They don't work and caused exceptions in my log.
The whole thing is running on java 11 instead of java 8 now.
High regards,
By Karn.