panik

Gracefully handle application downtimes and server errors

How it works

This project is intended to run alongside nginx or another reverse proxy. Configuration example:

error_page 500 502 503 504 /500.html;
location /500.html {
    root /srv/http;
}

location /panik/ {
    proxy_set_header X-App-Id "XptoApp";
    proxy_pass http://unix:/run/panik.socket:/app/;
}

With this setup, we can run a script from panik app on 500.html page, that checks application status, read logs and refreshes the browser when it comes back online:

<script src="/panik/init.js"></script>

Example webpage:

+------------------------------------+
|           + 5xx error +            |
|                                    |
|   /  Application is currently  \   |
|   \  restarting. Please wait.  /   |
|                                    |
|       Last online: 1 min ago       |
|       Logs: ~~~~~~~~~~~~~~~~       |
|             ~~~~~~~~~~~~~~~~       |
|             ~~~~~~~~~~~~~~~~       |
|                                    |
+------------------------------------+

Features / Requirements

• systemd (dbus) and systemd-journald integration
• (opt) docker and docker-compose integration
• websocket or wamp protocol?
• (opt) backoffice
• internal api to start / restart / shutdown applications
• check if application is healthy via http / systemd
• keep online status, last online, and other useful metrics (TBD)

Stack (TBD)

• (op1) python 3.7+, tornado, sqlite
• (op2) python 3.7+, django + channels, sqlite
• (op3) python 3.7+, autobahn, sqlite
• (op4) go, sqlite