zero_derivative

FWIW My boot delays were much more drastic and consistent across 6 Odroid C2's initially using the minimal image ... dmesg | grep -E "(rng|random)" [ 0.000000] random: get_random_bytes called from start_kernel+0xa8/0x400 with crng_init=0 [ 3.074316] random: fast init done [ 4.216928] random: systemd: uninitialized urandom read (16 bytes read) [ 4.220823] random: systemd: uninitialized urandom read (16 bytes read) [ 4.221479] random: systemd: uninitialized urandom read (16 bytes read) [ 1846.057161] random: crng init done [ 1846.057175] random: 7 urandom warning(s) missed due to ratelimiting My Odroid C2's also show a ridiculously low amount (single to low double digits) of available entropy without haveged and the rate at which available entropy increases is very very very slow ... cat /proc/sys/kernel/random/entropy_avail 5 With haveged it's marginally better, and the rate at which available entropy increases is better than without out it, but still too slow ... cat /proc/sys/kernel/random/entropy_avail 77 With rng-tools installed things are looking much better ... cat /proc/sys/kernel/random/entropy_avail 2002 and now SSH is available much quicker ... dmesg | grep -E "(rng|random)" [ 0.000000] random: get_random_bytes called from start_kernel+0xa8/0x400 with crng_init=0 [ 2.954319] random: fast init done [ 4.298379] random: systemd: uninitialized urandom read (16 bytes read) [ 4.308032] random: systemd: uninitialized urandom read (16 bytes read) [ 4.309349] random: systemd: uninitialized urandom read (16 bytes read) [ 10.320571] random: crng init done [ 10.320579] random: 7 urandom warning(s) missed due to ratelimiting Awesome! Thanks @guidol for the rng-tools tip

Greetings. I'm running a headless cluster of Odroid C2's (eMMC) using the latest Armbian buster minimal image with kernel 4.19.69-meson64. I'm encountering an issue where it takes a long time for SSH connection to be accepted (initially up to 30 min consistently). I get a connection refused. The issue seems to be due to the entropy pool becoming depleted during the early boot process which blocks SSH from starting while it refills the pool. The problem seems to be known and documented well here: https://daniel-lange.com/archives/152-Openssh-taking-minutes-to-become-available,-booting-takes-half-an-hour-...-because-your-server-waits-for-a-few-bytes-of-randomness.html As suggested I have installed installed haveged which brings the SSH startup from 30 mins to 10 mins after boot, which is an improvement, but still not good. Has anyone else experienced this issue? Any ideas?