Instance Hiccups
Instance Hiccups
When I first came to the Fediverse, having been chased away from Meta by censorship and witch hunts against Gnu/Linux enthusiasts, I finally checked out Mastodon. Something I had intended to do for a long time though never got around to. I just did not like the Mastodon Micro blogging format. I was never on Twitter for that precise reason. Though Mastodon was friendlier than Twitter, I still just did not like it. So I looked for other alternatives. I really like Friendica. It has its warts, make no mistake, I deal with them often. You do not even want to look at my error logs caused by the system itself. It would make a strong person stumble and weep. With that said, I still prefer it over other Instances I have reviewed which also produce plenty of errors themselves. So when I set out to setup my own Instance (3 - 5 days into my entering the Fediverse) I chose Friendica. I have not regretted that decision.
There have been a lot of challenges. Friendica's main development team uses Ubuntu server and I will not use anything from Canonical. I could have used a Debian Server though I prefer RHEL based distros. So I went with Rocky Linux 9, which added complexities to the install but undaunted I pressed forward and got my Instance up and running. I made some really bad choices to begin with. I installed via a zip file and then decided that was a really poor decision. It complicated upgrading. So I re-installed via git, which I could have just converted my install to a git repository but seeing I really had not launched yet it was easier and faster to just re-install. I then had a constant fight with SELinux to keep it from preventing upgrades and services I required from running. After I finally completely got SELinux configured I ran into database scheduling issues. Which I found a cure for on the Friendica GitHub Issues. One issue I have never managed to resolve was how php-fpm will just decide to die for some reason without a consistency that I could track down. So I put a band-aid on it. I wrote a script to check the services httpd, Mariadb, and php-fpm. If they are down then start them. This script is run every 5 minutes on the server. Until I can find what is killing php-fpm and fix that issue the band-aid will have to do. Sometimes the instance will run for days or weeks without incident, yes I wrote in logging aspects to the script, then other days it will crash several times a day. I also wrote a script to restart all services at midnight (My servers time PDT/PST) the refreshing has helped with some issues that come from running a service via PHP rather than having a proper daemon written in a system development language.
The big take from this discussion is that if the Instance is down when you try to login give it a few minutes and it will likely resolve itself. I have had a few times when the issue was with my VPS provider but they are very reliable so I see no reason for a change. In the future when I have more time to dedicate to the Instance, other than just keeping it up and running. I will chase down these bugs and make the changes necessary to avoid the need for band-aids. Though, that is probably not going to be in the near future. The Instance still has a great up time to down time ration compared to many other Instances I have seen. So I will keep on keeping on.
Unus Nemo
plan-A (゚ヮ゚) likes this.