Service log rate limiting doesn't work

[dtardon]

systemd version the issue has been seen with

current main (commit b2062ee)

Used distribution

Fedora 38

Linux kernel version used

6.7.9-100.fc38.x86_64

CPU architectures issue was seen on

x86_64

Component

systemd-journald

Expected behaviour you didn't see

Service won't log more than LogRateLimitBurst= messages during a given interval.

Unexpected behaviour you saw

Service logs all messages, regardless of rate-limiting settings.

Steps to reproduce the problem

cat > /tmp/verbose.awk <<EOF
#!/usr/bin/awk
BEGIN {
	i = 1
	for (j = 1; j <= 5; j++) {
		for (k = 1; k <= 50; k++)
			print "Here is startup message #", i++
	}
}
EOF

systemd-run --wait -u reproducer.service -p LogRateLimitBurst=200 -p LogRateLimitIntervalSec=10 awk -f /tmp/verbose.awk
journalctl -u reproducer.service --grep startup | wc -l

Additional program output to the terminal or log subsystem illustrating the issue

No response

[bluca]

Is this a regression or a pre-existing issue in 255?

[yuwata]

I guess this is expected. See below:

systemd/src/journal/journald-rate-limit.c

Lines 173 to 182 in 1bd0cef

	/*
	* Example:
	*
	* <= 1MB = rate * 1
	* 16MB = rate * 2
	* 256MB = rate * 3
	* 4GB = rate * 4
	* 64GB = rate * 5
	* 1TB = rate * 6
	*/

[dtardon]

Is this a regression or a pre-existing issue in 255?

No, not a regression.

[dtardon]

I guess this is expected. See below:

Hmm... systemd.exec(5) should be improved then, because it describes LogRateLimitBurst= as if it were a hard limit: If, in the time interval defined by LogRateLimitIntervalSec=, more messages than specified in LogRateLimitBurst= are logged by a service, all further messages within the interval are dropped until the interval is over.