A few days ago I decided to add a primitive logging feature to services so I would know when the services were actually started. At around 8:15pm EST MariaDB was started and at 9:15pm EST PHP-FPM was started. Starting the services when they go down does not fix the problem that causes their failure in the first place. Though it does keep the Instance up and running. One small victory.

#!/usr/bin/env bash

# Check to be sure all services are active, if not start them
# This shell script is invoked by cron every 15 minutes on this server
# note: The script does nothing if the services are active.

# Added primitive logging by appending to ../log/services.log every
# time that a service needs to be started.
# It is assumed that this script is launched from /var/www/html/bin 
# and services.log is in /var/www/html/log

apache_status=$(systemctl is-active httpd)
mariadb_status=$(systemctl is-active mariadb)
phpfpm_status=$(systemctl is-active php-fpm)

if [ "$mariadb_status" == "failed" ]; then
	systemctl start mariadb
        echo "$(date) Had to start MariaDB!" >> ../log/services.log
fi

if [ "$phpfpm_status" == "failed" ]; then
       systemctl start php-fpm
       echo "$(date) Had to start PHP-FPM!" >> ../log/services.log
fi

if [ "$apache_status" == "failed" ]; then
	systemctl start httpd
        echo "$(date) Had to start Apache!" >> ../log/services.log
fi

unset apache_status
unset mariadb_status
unset phpfpm_status

Updated Laptop Yesterday

Fedora 42 released and I went ahead and updated my laptop. I usually wait for a few months for them to work out the early release cycle bugs. 😉 Though I had been hearing good reviews from early adopters so I went ahead and took the plunge. I have been primarily using Fedora since F19. Canonical was pissing in my Wheaties again and I finally said to hell with them. I had tried Fedora before though I had drifted away looking for greener pastures. There are not a whole lot of General Purpose (GP) GNU/Linux distros that I have not tried at some point in time. Though my main 4 still remain:

• Fedora
• Arch
• LFS (Linux From Scratch)
• Alpine Linux

Of course I use Rocky Linux for my servers if possible though that is not a daily driver. I have been using GNU/Linux since 1993 when I heard about it on the UseNet and decided to download it, compile it and give it a whirl. It was not until 1997 or 1998 when Boot Magazine (now Wired) had Debian on a CD that I got to try out X Server. You see X Server is huge and on dial up that would be a several day adventure downloading so I had just put together a console. Not to say I had no graphics, It just wasn't on X. I have had to keep up with MS/DOS and then Windows because of work and family and friends though I have not had a MS Operating System on a machine of mine, except for testing, since 1993.

Why Fedora? Why Gnome? Well, the short and long of that is that it works for me. You see there is no best distro or DE (Desktop Environment). There is only the distro and DE that is best for you. So if someone is trying to shove their favorite distro down your throat, smile and check out some distros for yourself. Try the different DEs. We are fortunate enough to have many of them to choose from. Some OSes are stuck with, well, no choice.

The Instance was down twice last night. The first time I restarted MariaDB from my phone via a SSH session. The second time it was near time to go home when I noticed so I let it be until I could get to the house. I decided that seeing I cannot seem to get the services tweaked for all cases and I have not seen what is causing them to crash. Though PHP-FPM and MariaDB take turns shutting down on the Instance. I would at least like to make sure the Instance does not stay down for a prolonged period of time. I wrote a quick and dirty bash script to check the services on the Instance and to start them if they are not active. I then added a line on the system crontab to run the script every 15 minutes. So in a worse case scenario the Instance could be down for 15 minutes if it happens to go down directly after a check. For now this will make sure the Instance stays up until I can find out what is causing the shutdowns and fix the issues. I am sure there are many and it is not just one simple problem. 😉

If any other administrator is curios the script is quite simple.

#!/usr/bin/env bash

# Check to be sure all services are active, if not start them
# This shell script is invoked by cron every 15 minutes on this server
# note: The script does nothing if the services are active.

apache_status=$(systemctl is-active httpd)
mariadb_status=$(systemctl is-active mariadb)
phpfpm_status=$(systemctl is-active php-fpm)

if [ "$mariadb_status" == "failed" ]; then
	systemctl start mariadb
fi

if [ "$phpfpm_status" == "failed" ]; then
       systemctl start php-fpm
fi

if [ "$apache_status" == "failed" ]; then
	systemctl start httpd
fi

unset apache_status
unset mariadb_status
unset phpfpm_status

This of course only works if you are using systemd to manage your services on your system. I named the script services (no extension as bash scripts should not have an extension in most cases). I also invoke it in via cron every 15 minutes. The crontab is located in /etc/crontab for most 'nix type systems. I put the bash script in /var/www/html/bin where all the other Instance scripts are. You should have cron run the script as root so that the services can be started if need be.

# Example crontab line in /etc/crontab
# insert this at the end of the current crontab

*/15 * * * * root cd /var/www/html/bin; ./services

Sorry for the inconvenience of the Instance being down.

My work cycle varies. I work 12 hour shifts so on work days I do not do much more than Work, Sleep and do it again. This Week is a Long Week meaning that there are four days in the work cycle instead of 3. Though the way the Week is broken up I work from this Wednesday to next Sunday. Though Sunday is actually counted as one of the days on next weeks work cycle. So I work 5 consecutive days. Have 3 off, then work 2 days and have 4 off. That is what my work cycle breaks down to. During a Work cycle I do not have a lot of time to do much besides work and rest as already stated.

I will keep an eye on the server and make sure it stays up. I cannot do much more than that. Last night there was down time due to the MariaDB shutting down. This is likely yet more configuration Tuning.

If you setup up a LAMP or LEMP and install Friendica, turn on debugging the error logs go nuts. There are many known issues with the code base though 99.8% of them are non terminal. So they get worked around for now. I will eventually start refactoring code. Though in the mean time I have to keep the server up and running. My instance is Pristine that does not make it special it means that I have not modified any of the code base at all. I have only modified the settings. It is not actually a Vanilla installation as I do use Addons. So if Friendica works on other Instances then it will work here as well. Though it is my understanding that most Friendica Admins deal with the types of issues I am overcoming.

I will be around more after my work cycle. I hope everyone has a great week!

This is just a general announcement that I will be rebuilding the backend of the server that this Friendica Instance operates on within a month. I will give notice when I have a specific date.

Within the first week of being on The Fediverse I decided I was going to have my own Friendica Instance. I like Friendica more than any of the other Instance types I have ran into for many reasons. That does not mean it is exactly what I want it to be. I do plan many improvements as I have more time to work on them. For now I am just trying to get this Instance stable.

In my rush to get my Instance up I took some short cuts that I am now going to have to pay the price for. For instance my first install was via Tar Balls and I did not like the idea of how much work updates were going to be. Especially in applying needed patches between versions. So I re-installed via cloning the github project. This is always a bad idea. I should have forked the project first and then cloned from my own fork so that I could maintain my own branch in a more sane fashion. So I will be doing what should have been done in the first place had I spent more time planning an researching instead of diving in.

What this means to Instance Users is that when I do redo the backend the server may be down for a few hours possibly more. I will be letting you know ahead of time to backup your profiles. I will of course backup the database before I begin though I feel it is important that you have a chance to do backups of your own should you feel the need. Personally I feel you should. You never know when some catastrophic issue could cause data loss. So it is not a bad idea to do regular backups on things you feel are important in any case.

Take care, I hope you are enjoying your time on The Fediverse and I will be sure to keep everyone up to date with scheduled maintenance.