Moving to Centos 8.x with all that systemd crap.
Any advice?
Any advice?
;-)
On 12-08-20 21:54, Dennis Katsonis wrote to Gerrit Kuehn <=-
Systemd isn't that bad. It's much better than Lennart's pulseaudio,
now THAT sucks.
I moved to Systed with Fedora in 2013 or so, and really, the only four notable differences was the system booted a little faster, shut down a
lot faster, the commands to start a service were slightly different,
and the message log was accessed through journalctl.
On 12-08-20 13:16, Gerrit Kuehn wrote to Dennis Katsonis <=-
Hello Dennis!
08 Dec 20 21:54, Dennis Katsonis wrote to Gerrit Kuehn:
Systemd isn't that bad. It's much better than Lennart's pulseaudio,
now THAT sucks.
Tell me about it. OSS always worked fine for me. So did JACK.
As I said, this wholly depends on your use case. If you're running a notebook, I totally see the benefit. However, I have mainly servers and
a few workstations to maintain. These are not booted for days, weeks or even months, and they hardly ever change their network settings. If
they boot, hardware detection alone might take minutes, so I absolutely don't care about saving a few seconds afterwards during boot.
On the other hand, from day one I had a hard time making systemd *not* causing race conditions and actually wait for things that are required during boot on these machines, especially network connections and
remote mounts.
Systemd isn't that bad. It's much better than Lennart's pulseaudio, now THAT sucks.
Tell me about it. OSS always worked fine for me. So did JACK.
OSS worked, but limited in some ways. ALSA generally works well for
my needs.
JACK looks good, but I've never actually used it, even though it's
been around for donkeys years.
The --user option looks interesting. I might be able to use that
here if my init supports it.
There's another option, which allows running several binkd instancesand this crontab runs as root, oh dear :)
for different users:
% crontab -l | grep fidomailer
*/15 * * * * fidomailer
And, of course, it doesn't require the root privileges.i noted this in binkd.conf, but i miss the part of parse gentoo env
if its ignored to not use root in the --user openrc script
gentoo (openrc) do support multiple openrc users config deamons, no
need for bashing :)
Yes. Stop wasteing our time and let's do something useful.
For example teach me if it's wasteing or wasting? I do write looking
and watching but with "e" if it's boeing. What's the rule of that??
Actually fighting that idea of journalctl. OK, nice, new, complex, holistic...
There is usualy one need: to keep logs for certain amount of time
(based on SLA, based on law, etc.) Looks to me that design focused to zillion switches how to influence/keep disk space, but how to achieve
that I have 100% of time frame logged is not so easy. Shall I try to calculate disc space? Shall I rather put there all space to be sure?
From man: "...Normally, time-based deletion of old journal files
should not be required as size-based deletion with options such as SystemMaxUse= should be sufficient to ensure that journal files do not
grow without bounds..."
Look at logrotate which is the normal product to control journal files
in /var/log
Hello Vincent!
20 Dec 20 14:05, you wrote to me:
Look at logrotate which is the normal product to control journal
files in /var/log
The issue is that in default journalctl is not creating standard files
in /var/log/journal to be rotated as usual (which was my behaviour
before).
My understanding was that "vacuum" is done by journald itself.
But OK, I will figure it somehow...
I do not like using journalctl to read such logs so I allow a command
to run to use the older method and that way I can just look at an individual log file.
| Sysop: | Coz |
|---|---|
| Location: | Anoka, MN |
| Users: | 2 |
| Nodes: | 4 (0 / 4) |
| Uptime: | 188:17:01 |
| Calls: | 251 |
| Calls today: | 1 |
| Files: | 5,575 |
| Messages: | 225,044 |