Resource Consumption

[BerriJ]

Hey there,

first of all, thank you for this fantastic project. I enjoy using it. I set it up on a home server and noticed some spikes in the power consumption of this little device:

You can see that I started deploying it (with some troubles unrelated to solidtime) at ~18:00 o'cock. Around 20:45, I was finished and had this "spike-pattern" I had not had before.

Today, I investigated a bit and found that these occur every minute with a more pronounced spike every 10 minutes:

I quickly found that this is caused by this cron job:

solidtime/docker/prod/Dockerfile

Line 127 in 098ead8

&& echo "*/1 * * * * php ${ROOT}/artisan schedule:run --no-interaction" > /etc/supercronic/laravel

That runs each minute. I mitigated that by overwriting this file in my docker-compose.yml via a bind mount for the scheduler:

  scheduler:
    restart: always
    image: "solidtime/solidtime:${SOLIDTIME_IMAGE_TAG:-latest}"
    user: "1000:1000"
    # privileged: true
    networks:
      - internal
    volumes:
      - "app-storage:/var/www/html/storage"
      - "./logs:/var/www/html/storage/logs:Z"
      - "./app-storage:/var/www/html/storage/app:Z"
      - ./custom-laravel-cron:/etc/supercronic/laravel:ro

custom-laravel-cron is a text file just containing */15 * * * * php /var/www/html/artisan schedule:run --no-interaction. So I pushed the corn Job from 1 minute > 15 minutes. The result was already pretty good:

This removed the spikes every 60 seconds (yay), but now we still have smaller spikes every 30 seconds. These were caused by the health checks that each container runs.

Commenting out these healthcheck sections from my docker-compose.yml removed these remaining spikes:

Now I have a nice, resource-efficient deployment.

I just wanted to share this with others. However, it would be really nice to give the user more control over the job schedules (maybe by using variables in one of the .env files). Overwriting using a bind mount is not that obvious, and I don't know if this causes some unwanted side effects.

Lastly, do you know if these health checks are crucial in the compose file? If not, it may be a good Idea to comment them out or explain their use in the documents.

Again, thanks for this fantastic project.

[korridor]

Thanks for your comment. The self-hosting guide have the goal to provide a more or less production-ready setup.

Scheduler
Running the scheduler every minute is a normal approach for Laravel applications. It allows us developers to add/remove cronjobs in the applications code up to every minute. We currently don't have a cronjob that runs every minute, but changing that still might have consequences. We currently have a scheduled job that runs every 10 minutes and this one will likely not work properly now in your setup. If you are fine with that your solution is fine.
We can think about making this easier to customize, but to be honest this seams like this would creates more problems then it solves.

Healthchecks
You can remove the healthchecks and this will make your setup less resource intensive, but then you don't have healthchecks anymore. If you are fine with that then you can uncomment the lines in the docker-compose.yml. But I would advice to use healthchecks in a professional setup.

In summary: Removing stuff often has not only an effect on how many resources are being used, but on how stable or correct an application runs.