systemd: Yet Another Init Replacement

Cosmin · 2011-04-03 15:50:12

falconindy wrote:

Cosmin wrote:
I get this error when I install systemd
Failed to get D-Bus connection: Failed to connect to socket /org/freedesktop/systemd1/private: Connection refused
any ideas?
If you installed (not upgraded) just systemd, something in awry. Nothing in the install scriptlet could generate this error. Did you mean that this came from installing the initscripts package? In the end, you can ignore this error and reboot. It's harmless if systemd isn't even running (which is generally what causes the error).

I upgraded = reinstalled because It boot's fine but in gdm I have no user list and I can not log in ...

unikum · 2011-04-03 19:02:25

I try modify crond.service from Fedora 15 for Arch

[Unit]
Description=Cron Daemon
After=syslog.target

[Service]
EnvironmentFile=/etc/conf.d/crond
ExecStart=/usr/sbin/crond -f ${CROND_ARGS}
StandardOutput=syslog

[Install]
WantedBy=multi-user.target

Also other interesting service files
bluetooth.service

[Unit]
Description=Bluetooth Manager
After=syslog.target

[Service]
Type=dbus
BusName=org.bluez
ExecStart=/usr/sbin/bluetoothd -n
StandardOutput=syslog

[Install]
WantedBy=bluetooth.target

alsa-restore.service

[Unit]
Description=Restore Sound Card State
DefaultDependencies=no
After=sysinit.target
Before=shutdown.target
Conflicts=shutdown.target

[Service]
Type=oneshot
ExecStart=/usr/sbin/alsactl restore
StandardOutput=syslog

alsa-store.service

[Unit]
Description=Store Sound Card State
DefaultDependencies=no
Before=shutdown.target

[Service]
Type=oneshot
ExecStart=/usr/sbin/alsactl store
StandardOutput=syslog

laptop-mode.service

[Unit]
Description=Laptop power saving package
After=syslog.target acpid.service
Wants=acpid.service

[Service]
ExecStart=/usr/sbin/laptop_mode auto
ExecStop=/usr/sbin/laptop_mode stop

[Install]
WantedBy=multi-user.target

sensors.service

[Unit]
Description=Set Sensor Limits
DefaultDependencies=no
After=local-fs.target
Before=sysinit.target

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStart=-/usr/bin/sensors -s
ExecStart=-/usr/bin/sensors

[Install]
WantedBy=multi-user.target

I hope this will useful.

Last edited by unikum (2011-04-04 21:26:33)

Army · 2011-04-04 20:19:06

I just discovered that systemd is in [community], so I thought, this might be the right time to give it a shot and I'm quite amazed, especially by the speed!

Right now I only have two issues:
1. I want to start mpd on boot, so I have put it in the DAEMONS line in the rc.conf, but it fails to start. Every other daemon starts fine. Any idea what this might be caused by?
2. I want to achieve this https://wiki.archlinux.org/index.php/St … .2Finittab "auto-login" with systemd. I know it's really not good concerning security, but I'm ok with that. What exactly do I have to do to get this with systemd?

litemotiv · 2011-04-04 21:07:05

Army wrote:

2. I want to achieve this https://wiki.archlinux.org/index.php/St … .2Finittab "auto-login" with systemd. I know it's really not good concerning security, but I'm ok with that. What exactly do I have to do to get this with systemd?

See the previous page of this thread.

falconindy · 2011-04-04 21:16:15

Army wrote:

I just discovered that systemd is in [community], so I thought, this might be the right time to give it a shot and I'm quite amazed, especially by the speed!
Right now I only have two issues:
1. I want to start mpd on boot, so I have put it in the DAEMONS line in the rc.conf, but it fails to start. Every other daemon starts fine. Any idea what this might be caused by?
2. I want to achieve this https://wiki.archlinux.org/index.php/St … .2Finittab "auto-login" with systemd. I know it's really not good concerning security, but I'm ok with that. What exactly do I have to do to get this with systemd?

I do not encourage you to use the DAEMONS array in systemd. While there is some minimal support for through arch-daemons.target, I'd much rather you use the service file available in systemd-arch-units. Install that package and run:

# systemctl enable mpd.service
# systemctl start mpd.service

You can (should) remove the PID file entry from your mpd config.

falconindy · 2011-04-05 00:37:37

FYI: http://mailman.archlinux.org/pipermail/ … 19280.html

ngoonee · 2011-04-05 02:29:19

falconindy wrote:

FYI: http://mailman.archlinux.org/pipermail/ … 19280.html

This change does not affect those of us who ignore logs, does it =p

Army · 2011-04-05 08:25:07

Now this is strange, all of a sudden systemd doesn't boot anymore. Here's what I get:

[...]
systemd[1]: systemd-modules-load.service: main process exited, code=exited, status=1
Starting Load Kernel Modules failed, see 'systemctl status systemd-modules-load.service' for details.
systemd[1]: Unit systemd-modules-load.service entered failed state

Then it pauses for about 3 Minutes and continues with

Starting Load Random Seed
systemd[1]: Job dev-disk-by\...\....device/start timed out
Starting Recreate Volatile Files and Directories...
systemd[1]: Job dev-disk-by\...\....device/start timed out
systemd[1]: Job dev-tty2.device/start timed out
systemd[1]: Job dev-tty5.device/start timed out
systemd[1]: Job dev-tty1.device/start timed out
systemd[1]: Job dev-tty3.device/start timed out
systemd[1]: Job dev-tty4.device/start timed out
systemd[1]: Job dev-tty6.device/start timed out
Starting Permit User Sessions
systemd[1]: Startup finished in 4s 94ms 938us (kernel) + 3min 2s 35ms 104us (userspace) = 3min 6s 130ms 42us

and then it stops. I didn't do any customizations, just did a reboot. I tried to reinstall everything systemd needs, before that I deleted everything the install scripts created and didn't remove after deinstallation. Doesn't help, it doesn't work anymore...

Any ideas?

falconindy · 2011-04-05 10:12:06

Important detail you've left out -- what version?

Army · 2011-04-05 10:32:00

Oh, I started using systemd 22-2, now there's a newer version in [community-testing], 23-1. And the other packages are initscripts-systemd v21-1 and systemd-arch-units 20110320-2. The issue stays the same with newer and old systemd. That's why I didn't post this info, because I think that's not relevant. Or is it?

Last edited by Army (2011-04-05 10:33:56)

falconindy · 2011-04-05 12:51:39

It's always relevant, since systemd is moving along fairly quickly. I was initially suspicious of v23 running with udev 166, but quickly squashed that idea on my VM. If you up the log verbosity (add 'verbose' and 'systemd.loglevel=debug' as boot params), you'll get a better idea of what's actually hanging.

Army · 2011-04-05 13:40:26

Ok thanks, I'll try these asap.

ngoonee · 2011-04-05 14:12:13

Is there some sort of minimum kernel version for systemd? I have an old 2.6.33 kernel lying around on my system which I decided to try and boot. Booted fine but I couldn't connect to dbus (thus no systemctl etc.). Not a big deal, am not using that kernel at all, but since I've never heard of a minimum kernel (or even any sort of kernel dependency) I thought it odd.

falconindy · 2011-04-05 14:18:57

2.6.33 isn't going to work out too well. At a minimum, you need 2.6.36 for the cgroup API FS mounted at /sys/fs/cgroup, and CONFIG_DEVTMPFS enabled (2.6.37 with fanotify is optional for readahead support). And of course, the latest changes bumping that up to 2.6.39 unless we hear of plans to backport.

edit: touched the wiki to note the 2.6.36 minimum.

Last edited by falconindy (2011-04-05 14:31:34)

Army · 2011-04-05 15:06:11

How can I get this output from the boot process into a file? That would help a lot, because I have to show it to you guys, I can't figure out what causes the problem here.

falconindy · 2011-04-05 16:37:38

syslog should be capturing it. If it's not, you might need socket activated syslog-ng to extract from the kmsg buffer

http://aur.archlinux.org/packages.php?ID=46340

ngoonee · 2011-04-06 00:58:22

falconindy wrote:

2.6.33 isn't going to work out too well. At a minimum, you need 2.6.36 for the cgroup API FS mounted at /sys/fs/cgroup, and CONFIG_DEVTMPFS enabled (2.6.37 with fanotify is optional for readahead support). And of course, the latest changes bumping that up to 2.6.39 unless we hear of plans to backport.
edit: touched the wiki to note the 2.6.36 minimum.

Thanks, figured it'd be something like that. Oh well, no problem using sysvinit on that kernel's grub-line after all . Thanks.

eworm · 2011-04-06 08:34:14

eworm wrote:

[...] has some problems on Thinkpad Notebooks (T400s and T410s). It basically works, but I have a boot delay of several minutes. [...]

This issue is solved with udev-167 (built the package myself) and systemd-23. The unit udev-settle.service is no longer started by default, so it can not cause any delays.

litemotiv · 2011-04-06 08:46:27

I notice that each time i suspend/resume, a number of mounts are duplicated:

% mount

proc on /proc type proc (rw,nosuid,nodev,noexec,relatime)
sys on /sys type sysfs (rw,nosuid,nodev,noexec,relatime)
udev on /dev type devtmpfs (rw,nosuid,relatime,size=10240k,nr_inodes=472716,mode=755)
/dev/disk/by-uuid/4bb96692-5139-4c79-b60f-2d181412b8bd on / type ext4 (rw,noatime,discard)
shm on /dev/shm type tmpfs (rw,nosuid,nodev)
debugfs on /sys/kernel/debug type debugfs (rw)
securityfs on /sys/kernel/security type securityfs (rw)
mqueue on /dev/mqueue type mqueue (rw)
hugetlbfs on /dev/hugepages type hugetlbfs (rw)
binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)
/dev/sda4 on /home type ext4 (rw,noatime,discard)
debugfs on /sys/kernel/debug type debugfs (rw)
securityfs on /sys/kernel/security type securityfs (rw)
mqueue on /dev/mqueue type mqueue (rw)
hugetlbfs on /dev/hugepages type hugetlbfs (rw)
binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)
fusectl on /sys/fs/fuse/connections type fusectl (rw)
/dev/sda4 on /home type ext4 (rw,noatime,discard)
debugfs on /sys/kernel/debug type debugfs (rw)
securityfs on /sys/kernel/security type securityfs (rw)
mqueue on /dev/mqueue type mqueue (rw)
hugetlbfs on /dev/hugepages type hugetlbfs (rw)
binfmt_misc on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)

Is this a systemd issue or something else?

falconindy · 2011-04-06 10:12:17

Neat -- easy enough to check if its systemd by booting without the init=/bin/systemd param.

litemotiv · 2011-04-06 10:20:31

falconindy wrote:

Neat -- easy enough to check if its systemd by booting without the init=/bin/systemd param.

Oh yea sorry, i forgot that i already did that, it only happens with systemd running.

stefano.facchini · 2011-04-06 14:15:56

Hi,
I'm trying to mount /usr as a squashfs, as explained in https://wiki.archlinux.org/index.php/Ma … ing_.2Fusr , but during the boot systemd produces errors when mounting the squashed filesystem. Maybe adding lines to /etc/fstab does not work well for systemd?

falconindy · 2011-04-06 23:39:43

stefano.facchini wrote:

Hi,
I'm trying to mount /usr as a squashfs, as explained in https://wiki.archlinux.org/index.php/Ma … ing_.2Fusr , but during the boot systemd produces errors when mounting the squashed filesystem. Maybe adding lines to /etc/fstab does not work well for systemd?

two things:
1) systemd does not support /usr on a separate partition. and my two cents: squashing directories like this is just a bad idea in general.
2) lines in /etc/fstab generate units (you'll see a bunch of .mount units via systemctl).

Last edited by falconindy (2011-04-06 23:40:46)

yoshi314 · 2011-04-07 08:28:41

is there a good way to capture systemd debug output into a file, if i cannot get to a login prompt?

it seems to lock up at processing udev events after a while.

eworm · 2011-04-07 08:54:03

yoshi314 wrote:

is there a good way to capture systemd debug output into a file, if i cannot get to a login prompt?

Logs should go to syslog.

yoshi314 wrote:

it seems to lock up at processing udev events after a while.

This is true for current udev from [core] on some systems. I had the same issue on my Thinkpad T410s.
You have two possible solutions for this:

add "--timeout=5" to /lib/systemd/system/udev-settle.service, this reduces the delay from 180 to 5 seconds
install udev-167 (compile from abs (just change version numbers and regenerate checksums) or use my version (for x86_64))

Or just wait for udev-167 to appear in the official repos.

Last edited by eworm (2011-04-07 08:54:24)

Arch Linux

#351 2011-04-03 15:50:12

Re: systemd: Yet Another Init Replacement

#352 2011-04-03 19:02:25

Re: systemd: Yet Another Init Replacement

#353 2011-04-04 20:19:06

Re: systemd: Yet Another Init Replacement

#354 2011-04-04 21:07:05

Re: systemd: Yet Another Init Replacement

#355 2011-04-04 21:16:15

Re: systemd: Yet Another Init Replacement

#356 2011-04-05 00:37:37

Re: systemd: Yet Another Init Replacement

#357 2011-04-05 02:29:19

Re: systemd: Yet Another Init Replacement

#358 2011-04-05 08:25:07

Re: systemd: Yet Another Init Replacement

#359 2011-04-05 10:12:06

Re: systemd: Yet Another Init Replacement

#360 2011-04-05 10:32:00

Re: systemd: Yet Another Init Replacement

#361 2011-04-05 12:51:39

Re: systemd: Yet Another Init Replacement

#362 2011-04-05 13:40:26

Re: systemd: Yet Another Init Replacement

#363 2011-04-05 14:12:13

Re: systemd: Yet Another Init Replacement

#364 2011-04-05 14:18:57

Re: systemd: Yet Another Init Replacement

#365 2011-04-05 15:06:11

Re: systemd: Yet Another Init Replacement

#366 2011-04-05 16:37:38

Re: systemd: Yet Another Init Replacement

#367 2011-04-06 00:58:22

Re: systemd: Yet Another Init Replacement

#368 2011-04-06 08:34:14

Re: systemd: Yet Another Init Replacement

#369 2011-04-06 08:46:27

Re: systemd: Yet Another Init Replacement

#370 2011-04-06 10:12:17

Re: systemd: Yet Another Init Replacement

#371 2011-04-06 10:20:31

Re: systemd: Yet Another Init Replacement

#372 2011-04-06 14:15:56

Re: systemd: Yet Another Init Replacement

#373 2011-04-06 23:39:43

Re: systemd: Yet Another Init Replacement

#374 2011-04-07 08:28:41

Re: systemd: Yet Another Init Replacement

#375 2011-04-07 08:54:03

Re: systemd: Yet Another Init Replacement

Board footer