@cstross @marjolica

Strictly speaking, it's DJB who's the newbie. Taking that long to come up with cyclog. (-:

In the early 1990s, work had taken me away from Unices to the world of PCs and compatibles.

(It's called multilog now, for anyone following along who might be scratching xyr head. Yes, multilog was 21st century. DJB's first iteration was cyclog, though.)

http://jdebp.info/Softwares/djbwares/guide/commands/multilog.xml

#DanielJBernstein #cyclog #s6

@cstross @marjolica

You've both made a rather large and quite amusing assumption, there, though.

I'm doing this with tooling invented by DJB in the 1990s, on OpenBSD 5.

I can _easily_ outdo _both_ of you on the not using systemd or Linux front. (-:

But I didn't mention that until now because this isn't about that. This is about the several suggestions in this thread to start using logrotate, which was bettered long before systemd was even thought of.

#OpenBSD #cyclog

@cstross @marjolica

Not mine, and not post-2010. Invented by Daniel J. Bernstein in the 1990s. That's low long logrotate has been passé. (-:

#cyclog

@cstross @motomatters

Not mine, and it's been on Hacker News numerous times, too, and others from StackExchange to Wikipedia.

Logging is strictly size capped through tooling that does this, properly. Bloating is not possible. I lose an old log file from 2016 to make way for new entries in 2023 when the size cap is reached. (That's how far back the http4d server logs happen to go right now.)

#cyclog

@motomatters @cstross

Another perspective:

I didn't have any access to the machine running my #GOPHER server for almost 3 years; but I had it configured with properly size-capped logs. (logrotate does not in fact guarantee strict size capping.)

I finally got some of the access back, this week. It's been quietly soldiering on since COVID-19, unattended.

JdeBP ~ $du -sh /var/log/sv/gopher4d/
22.4M /var/log/sv/gopher4d/
JdeBP ~ $

/var/log is not on the root volume, of course.

#cyclog