getrandom() initialization failed if run on boot

[laurentdinclaux]

Steps to reproduce the problem:

I created a systemd startup file:

[Unit]
Description=Mitmproxy
After=network.target

[Service]
#Type=simple
Type=idle
EnvironmentFile=/etc/default/mitmproxy
User=mitmproxy
Group=mitmproxy
LimitNOFILE=32768
ExecStart=/usr/local/bin/mitmdump $DAEMON_ARGS --conf=$CONFFILE --cadir=/etc/mitmproxy

[Install]
WantedBy=multi-user.target

Whenever I reboot my ubuntu 16.04 VM, mitmdump runs, but throw error for SSL websites:

Cannot establish TLS with 216.58.204.104:443 (sni: ssl.google-analytics.com): TlsException("SSL handshake error: Error([('', 'osrandom_rand_bytes', 'getrandom() initialization failed.')],)",)

If I log into ssh and restart the service manually it then runs fine.

System information

Ubuntu 16.04 - mitmproxy 1.0.2

[Kriechi]

This is more a general Python issue and unfortunately not directly related to mitmproxy.
Fur further information please see:
https://lwn.net/Articles/693189/
https://bugs.python.org/issue26839

You could try and experiment with systemd-random-seed.service in your service definition...

[nesttle]

Same here,
https://discourse.mitmproxy.org/t/run-mitmdump-from-crontab-solved/373/4

I try with:
@reboot sleep 60 && script.sh

Still not working (didn't happen on mitm0.18, but 2.0)