I can see there are some rate limits in /etc/systemd/journald.conf and the default values, according to journald.conf(5) is "1000 messages in 30s". This is nice. However, it seems that these limits do not apply to kernel messages.

My / partition is 10 GiB so journals by defaults take up to 1 GiB space, this is enough for months of logs on my laptop. However, yesterday I encountered a bug related to USB kernel module while traferring data from my laptop to my USB device, this bug made the USB kernel module keep generating kernel messages (that kind of [---cut here---] stuff) at a rate of thousands of lines per second. These kernel messages quickly took up 1 GiB space and totally flushed all the logs of previous months in a few minutes.

Generating more than 1 GiB of kernel messages in a few minutes is apprently more than the defaults limit of "1000 messages in 30s", but journald did not stop this kind of "stupid" behaviour.

May I suggest that journald should also apply the rate limit not only on service messages, but also on kernel messages? This will help a lot when some kernel module generates tons of messages.

Thanks.