systemd: Yet Another Init Replacement

bobwya · 2012-04-06 17:22:21

lucke wrote:

systemd in extra has that patch applied.

Ta, Just installed the updates - so I'm all good again... I'm not quite ready to start getting to grips with ABS (previous attempts have failed badly )...

Bob

eworm · 2012-04-06 18:41:52

I fixed my symlinks with this command:

for LINK in `find -L -type l`; do ln -sf /usr/`readlink $LINK` $LINK; done

(Use at your own risk... But it worked for me.)

falconindy · 2012-04-06 19:31:34

eworm wrote:

I fixed my symlinks with this command:
for LINK in `find -L -type l`; do ln -sf /usr/`readlink $LINK` $LINK; done
(Use at your own risk... But it worked for me.)

Please don't do this... systemctl has a 'reenable' verb that will do this for you.

eworm · 2012-04-06 21:03:28

Did not know that. Thanks for the hint!
Can I break anything linking manually?

ron9 · 2012-04-06 22:59:33

falconindy wrote:

Even "broken" symlinks from enabled services in /etc/systemd/system will still continue to function

Yes I saw avahi-daemon.service unit was started from rc.d after it was removed from /lib/systemd. Reminds me to remove rc.d q:-)

After dbus-core update to day the dbus unit files was recreated in /lib/systemd after I yesterday moved them to /usr/lib/systemd......

And dbus.socket and service files have no install section so I manually symlinked dbus.socket to multi-user.target.

bobwya · 2012-04-07 01:04:26

Hi all,

Could someone please kindly test out this (now working!!) patch for the systemd bash-completion script.

In the most recent versions of bash-completion package (extra/bash-completion 1.99-2) and systemd (extra/systemd 44-5) this will now be found at:

/usr/share/bash-completion/completions/systemctl

This code is more robust than the original version and correctly handles unit names with escape sequences
e.g. CIFS mounted at "/mnt/user-server/TV Series/A..M" = systemd automount unit file mnt-user\x2dserver-TV\x20Series-A..M.automount
will break current bash-completion script version.

Also what is the upstream for this script: is it Fedora or is this an ARCH creation? I don't see the systemctl file in the upstream bash-completion package at Debain...

Bob

falconindy · 2012-04-07 02:55:14

bobwya wrote:

Hi all,
Could someone please kindly test out this (now working!!) patch for the systemd bash-completion script.
In the most recent versions of bash-completion package (extra/bash-completion 1.99-2) and systemd (extra/systemd 44-5) this will now be found at:
/usr/share/bash-completion/completions/systemctl
This code is more robust than the original version and correctly handles unit names with escape sequences
e.g. CIFS mounted at "/mnt/user-server/TV Series/A..M" = systemd automount unit file mnt-user\x2dserver-TV\x20Series-A..M.automount
will break current bash-completion script version.
Also what is the upstream for this script: is it Fedora or is this an ARCH creation? I don't see the systemctl file in the upstream bash-completion package at Debain...
Bob

bash completion for systemd is shipped with systemd (pacman -Qo would have told you this). It's not called systemctl as shipped from upstream. It's mangled (and symlinked) to cope with the way bash-completion 1.99 does on-demand loading of completions.

bobwya · 2012-04-07 08:46:23

falconindy wrote:

bash completion for systemd is shipped with systemd (pacman -Qo would have told you this). It's not called systemctl as shipped from upstream. It's mangled (and symlinked) to cope with the way bash-completion 1.99 does on-demand loading of completions.

Dude,

I'm well aware that systemctl is part of the systemd package. I was merely pointing out that the default location for the bash-completion systemd script has recently changed from /etc/bash_completion.d to /usr/share/bash-completion/completions - in the case the patch is useful to anyone or they won't to test it. I know all this because of the hours I spent pouring over BASH scripting pages (since I am not too familiar with the shell language) and sweat to produce a working fix. I guess I won't bother doing this in the future...

Meh,
Bob

crabman · 2012-04-07 13:37:47

I have added some information about systemd-journal in conjunction with classic logging daemons to the the systemd wiki page:
https://wiki.archlinux.org/index.php/Sy … md_Journal

I also found the information in the official arch-announcement regarding the correct socket to be incorrect. According to lennart, the correct socket for syslogs to listen on is /run/systemd/journal/syslog and not /run/systemd/journal/socket (see http://lwn.net/Articles/474968/)

However, the creation of this socket didnt work for me, so i had to alter syslog-ng's .service file as mentioned here: http://mailman.archlinux.org/pipermail/ … 25082.html

So my questions are:
Was that really an error in the announcement or did i just get something wrong?
Is this a bug in the syslog-ng package or am i wrong again?

crabman

ron9 · 2012-04-07 22:50:13

When I do "enable" and then try to "start" a unit file I get:

[ronny@me ~]$ sudo systemctl start avahi-daemon.service
Failed to issue method call: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount' for details.

And "status" of mount give this:

[ronny@me ~]$ sudo systemctl status -.mount
systemctl: invalid option -- '.'

Journal log:

Apr 07 23:28:32 me systemd[1]: Cannot add dependency job for unit local-fs.target, ignoring: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount' for details.
Apr 07 23:28:32 me systemd[1]: Cannot add dependency job for unit proc-sys-fs-binfmt_misc.automount, ignoring: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount
Apr 07 23:28:32 me systemd[1]: Cannot add dependency job for unit systemd-ask-password-console.path, ignoring: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount

?

csn · 2012-04-09 09:27:05

ron9 wrote:

When I do "enable" and then try to "start" a unit file I get:

[ronny@me ~]$ sudo systemctl start avahi-daemon.service
Failed to issue method call: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount' for details.

And "status" of mount give this:

[ronny@me ~]$ sudo systemctl status -.mount
systemctl: invalid option -- '.'

Journal log:

Apr 07 23:28:32 me systemd[1]: Cannot add dependency job for unit local-fs.target, ignoring: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount' for details.
Apr 07 23:28:32 me systemd[1]: Cannot add dependency job for unit proc-sys-fs-binfmt_misc.automount, ignoring: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount
Apr 07 23:28:32 me systemd[1]: Cannot add dependency job for unit systemd-ask-password-console.path, ignoring: Unit -.mount failed to load: No such file or directory. See system logs and 'systemctl status -.mount

?

Hi,

Unfortunately `-.mount' (the root file system) cannot be referenced by that name from systemctl or from unit files since it doesn't follow the usual escaping rules of systemd: all the other mount units get their leading slash removed during name conversion, but since the root fs consists of only one slash it gets converted to a single dash instead. Otherwise it would consist of a suffix only which obviously can't work.

So, not knowing your specific (unit-)configuration my guesses would be:
a) you've tried to reference -.mount somewhere in a unit file which doesn't work (see above)
b) there might be something wrong with your root filesystem (e.g. it wasn't correctly remounted during startup or there's some kind of error in your /etc/fstab)

Maybe this can point you in the right direction.

Last edited by csn (2012-04-09 09:28:54)

falconindy · 2012-04-09 10:05:57

Yes, -.mount is indeed an invalid option. So tell systemctl that it isn't an option....

$ systemctl status -- -.mount
-.mount - /
	  Loaded: loaded
	  Active: active (mounted) since Sun, 08 Apr 2012 21:53:05 -0400; 8h ago
	   Where: /
	    What: /dev/sda2
	  CGroup: name=systemd:/system/-.mount

Seems to me like you're missing the "/" entry from /etc/fstab.

Maki · 2012-04-09 12:34:52

I have trouble setting up samba, both smbd.service and nmbd.service fail to start, while the wrapper around rc.d samba.service works just fine.

smbd.service - Samba SMB/CIFS server
          Loaded: loaded (/usr/lib/systemd/system/smbd.service; enabled)
          Active: failed (Result: exit-code) since Mon, 09 Apr 2012 14:26:39 +0200; 5min ago
         Process: 377 ExecStart=/usr/sbin/smbd -F (code=exited, status=1/FAILURE)
          CGroup: name=systemd:/system/smbd.service

nmbd.service - Samba NetBIOS name server
          Loaded: loaded (/usr/lib/systemd/system/nmbd.service; enabled)
          Active: failed (Result: exit-code) since Mon, 09 Apr 2012 14:26:40 +0200; 7min ago
         Process: 378 ExecStart=/usr/sbin/nmbd -F (code=exited, status=1/FAILURE)
          CGroup: name=systemd:/system/nmbd.service

Fixed the problem, the folder /run/samba didn't exist, thanks falconindy, also chek this thread https://github.com/falconindy/systemd-a … /issues/58

Last edited by Maki (2012-04-09 15:17:34)

ron9 · 2012-04-09 13:18:42

My laptop starts fine with no errors, and has worked perfectly with native systemd for months.

-.mount        loaded active mounted       /
boot.mount     loaded active mounted       /boot

But lately when I do systemctl enable "any" unit file, mount gives an error message, but it is still mounted.

-.mount        error  active mounted       /
boot.mount     loaded active mounted       /boot

Thanks to falconindy for showing systemctl who's the boss.

[ronny@me ~]$ systemctl status -- -.mount
-.mount - /
	  Loaded: error (Reason: No such file or directory)
	  Active: active (mounted) since Mon, 09 Apr 2012 08:04:11 +0000; 5h 29min ago
	   Where: /
	    What: /dev/sda3
	  CGroup: name=systemd:/system/-.mount

I have not made any changes to the fstab in the past. So what may have been wrong with my system...

# /etc/fstab: static file system information
#
# <file system> <dir> <type> <options> <dump> <pass>
tmpfs /tmp tmpfs nodev,nosuid 0 0
LABEL=boot /boot ext4 defaults,noatime,noauto,comment=systemd.automount  0 2
LABEL=arch / ext4 defaults,noatime 0 1
LABEL=mydata /home/ronny/mydata ntfs-3g noatime,uid=1000,gid=users,noauto,comment=systemd.automount 0 0

@csn It was silly of me, what I meant was that the output from systemctl wasn't correct.

See system logs and 'systemctl status -.mount' for details.

This output should be corrected in systemd.

unikum · 2012-04-09 16:19:59

`Hi.

Tell me please what services I should start for nfs server and nfs client.

UPD: seems in this thread should not expect help. Answer to myself:
For server

rpcbind.service rpc-mountd.service exportfs.service

For client

rpcbind.service rpc-statd.service

Last edited by unikum (2012-04-11 08:21:47)

mwc · 2012-04-11 00:41:58

Updated systemd-git today and audio stopped working after reboot due to a lack of ACLs on /dev/snd/*.

loginctl reports zero sessions and systemd-cgls shows all user processes in the gdm.service control group.

This was traced to an issue with the pam_systemd.so module. From the /var/log/gdm/:0-slave.log file:

gdm-password][1837]: PAM unable to dlopen(/lib/security/pam_systemd.so): /lib/security/pam_systemd.so: undefined symbol: audit_loginuid_from_pid
gdm-password][1837]: PAM adding faulty module: /lib/security/pam_systemd.so
gdm-password][1837]: pam_unix(gdm-password:auth): authentication failure; logname= uid=0 euid=0 tty=:0 ruser= rhost=  user=****

Any ideas besides watching gitweb?

graysky · 2012-04-11 00:47:57

@falconindy - naive question, but I couldn't answer it on the systemd wiki page: when I restart or shutdown the box via the xfce4 GUI, is it calling the proper "systemctl command" to gracefully stop all running daemons?

king.flasher.dave · 2012-04-11 13:14:55

android_808 wrote:

After the recent update, all symlinks in /etc/ have been removed for the graphical.target.wants. GDM readded fine but network manger doesn't seem to be there. I can't find any trace under /usr/lib. Trying to start nm-applet results in:
WARNING **: Failed to register as an agent: (32) Unit dbus-org.freedesktop.NetworkManager.service failed to load: No such file or directory. See system logs and 'systemctl status dbus-org.freedesktop.NetworkManager.service' for details.
Edit: Solved. Managed to get it to enable. Not quite sure what went wrong there.

How did you do that? my network.service still fails on startup. I have tried reenabling and whatnot, so far all I can do is manually restart the service after login to make it work... quite annoying

Hspasta · 2012-04-11 23:26:14

Does systemd mount my /media with tmpfs? Also, what are these run types that are being used for /var/lock and /var/run? I used to have tmpfs for them and used realized its changed.

tomegun · 2012-04-12 00:29:14

Hspasta wrote:

Does systemd mount my /media with tmpfs? Also, what are these run types that are being used for /var/lock and /var/run? I used to have tmpfs for them and used realized its changed.

systemd mounts a tmpfs on /media. However, this will no longer be the case as of the next release (as udisks2 does not use /media the folder can mostly be ignorde).

systemd prefers /var/run and /var/lock to be symlinks to /run (which is a tmpfs) and /run/lock, respectively. If this is not the case, they will be bindmounted instead. In both cases the contents of /var/run and /run will be identical.

65kid · 2012-04-13 16:35:53

I can't get my encrypted swap device to work with systemd. At first it asked me for a passphrase for the swap device on boot which it obviously shouldn't do. I fixed this by using /dev/urandom instead of the SWAP keyword in /etc/crypttab.
But I still can't boot. Now it just sits there with a blinking cursor and does nothing and I all I can do is reboot with Ctrl+Alt+Del.
The last messages are:

[...]
Starting /boot...
Started /boot [OK]
Starting Recreate Volatile Files and Directories...
Starting Load Random Seed...
Started Load Random Seed [OK]
Starting Cryptography Setup for swap...
Started Recreate Volatile Files and Directories [OK]
Started Cryptography Setup for swap [OK]

/etc/crypttab

swap		/dev/sda2		/dev/urandom		-c aes-cbc-essiv:sha256 -s 256

/etc/fstab

tmpfs			/tmp	tmpfs	nodev,nosuid	0	0
/dev/sda1		/boot	ext4	defaults	0	1
/dev/mapper/swap	swap	swap	defaults	0	0
/dev/mapper/muscrypt	/	ext4	defaults	0	1

I boot with:

root=/dev/mapper/muscrypt cryptdevice=/dev/sda3:muscrypt ro verbose init=/usr/lib/systemd/systemd

If I remove /dev/mapper/swap from /etc/fstab I can boot just fine.

edit:
nevermind, this fixed it:

swap		/dev/sda2		/dev/urandom		swap

Last edited by 65kid (2012-04-13 18:57:35)

graysky · 2012-04-14 11:06:04

ajk wrote:

Did anyone try systemd allready on their Arch Linux system? I just build v4 but the network still doesn't come up by default.. Anyone has a clue on how to fix this?

You read the wiki?

65kid · 2012-04-15 12:38:35

I'm a little confused about the automount option in /etc/fstab. Using noauto,comment=systemd.automount has been working fine for me. Now in the wiki and the systemd.mount man page it says to use x-systemd.automount instead. This didnt work at all, systemd didn't create an automount unit. Looking at the source the option actually seems to be x-systemd-automount (notice the -). So it seems like the man page is wrong here.

But using noauto,x-systemd-automount seems to be broken as well. The automount unit is created but accessing the mount point fails.

$ ls -l /boot
ls: cannot access /boot: No such device

$ systemctl status boot.mount
boot.mount - /boot
	  Loaded: loaded
	  Active: inactive (dead)
	   Where: /boot
	    What: /dev/sda1
	  CGroup: name=systemd:/system/boot.mount

Apr 15 14:27:36 mus mount[310]: mount: wrong fs type, bad option, bad superblock on /dev/sda1,
Apr 15 14:27:36 mus mount[310]: missing codepage or helper program, or other error

When I change back to noauto,comment=systemd.automount and reboot, everything works fine. I can also confirm the same behaviour on my RAID device.

algorythm · 2012-04-15 19:22:03

Nice. With readahead my boot speeds up by 6 seconds (from GRUB2 to desktop with all startup applications open), meaning 24s became 18.

Otherwise it's just as fast as the default init, despite its "aggressive parallelization capabilities" and all that shit, which was a bit disappointing.

graysky · 2012-04-15 19:52:56

algorythm wrote:

Otherwise it's just as fast as the default init, despite its "aggressive parallelization capabilities" and all that shit, which was a bit disappointing.

Faster on an SSD.

Last edited by graysky (2012-04-15 19:53:14)

Arch Linux

#1326 2012-04-06 17:22:21

Re: systemd: Yet Another Init Replacement

#1327 2012-04-06 18:41:52

Re: systemd: Yet Another Init Replacement

#1328 2012-04-06 19:31:34

Re: systemd: Yet Another Init Replacement

#1329 2012-04-06 21:03:28

Re: systemd: Yet Another Init Replacement

#1330 2012-04-06 22:59:33

Re: systemd: Yet Another Init Replacement

#1331 2012-04-07 01:04:26

Re: systemd: Yet Another Init Replacement

#1332 2012-04-07 02:55:14

Re: systemd: Yet Another Init Replacement

#1333 2012-04-07 08:46:23

Re: systemd: Yet Another Init Replacement

#1334 2012-04-07 13:37:47

Re: systemd: Yet Another Init Replacement

#1335 2012-04-07 22:50:13

Re: systemd: Yet Another Init Replacement

#1336 2012-04-09 09:27:05

Re: systemd: Yet Another Init Replacement

#1337 2012-04-09 10:05:57

Re: systemd: Yet Another Init Replacement

#1338 2012-04-09 12:34:52

Re: systemd: Yet Another Init Replacement

#1339 2012-04-09 13:18:42

Re: systemd: Yet Another Init Replacement

#1340 2012-04-09 16:19:59

Re: systemd: Yet Another Init Replacement

#1341 2012-04-11 00:41:58

Re: systemd: Yet Another Init Replacement

#1342 2012-04-11 00:47:57

Re: systemd: Yet Another Init Replacement

#1343 2012-04-11 13:14:55

Re: systemd: Yet Another Init Replacement

#1344 2012-04-11 23:26:14

Re: systemd: Yet Another Init Replacement

#1345 2012-04-12 00:29:14

Re: systemd: Yet Another Init Replacement

#1346 2012-04-13 16:35:53

Re: systemd: Yet Another Init Replacement

#1347 2012-04-14 11:06:04

Re: systemd: Yet Another Init Replacement

#1348 2012-04-15 12:38:35

Re: systemd: Yet Another Init Replacement

#1349 2012-04-15 19:22:03

Re: systemd: Yet Another Init Replacement

#1350 2012-04-15 19:52:56

Re: systemd: Yet Another Init Replacement

Board footer