systemd-journald discards messages from a service exceeding its burst limit and notes the fact in a single line that is easy to miss — so a debugging session goes quiet exactly when the process is busiest.
# /etc/systemd/journald.conf
RateLimitIntervalSec=30s
RateLimitBurst=10000
# or per unit, for one chatty service
# LogRateLimitIntervalSec=0The default is 1000 messages in 30 seconds, which a worker processing a queue exceeds without trying. The giveaway is Suppressed N messages from /system.slice/... in the journal. Raising it is fine on a box with disk to spare; the better answer for anything genuinely high-volume is to ship the logs from the application rather than routing them through the journal at all.