systemd: Yet Another Init Replacement

falconindy · 2011-08-22 22:00:22

yeah, that's normal. tagging is done by a udev rule which doesn't go away when systemd isn't PID 1, so it makes sense that the alias and tag still show up under sysvinit. As far as I can tell, the only difference is the final line of the systemd udev rule which runs systemd-sysctl. In a nutshell, this applies sysctl values to any subtree matching the network device triggering the udev rule. Rules are sourced from... (pardon my lazy copy/paste, im on a bus)

"/run/sysctl.d",
"/etc/sysctl.d",
"/usr/local/lib/sysctl.d",
"/usr/lib/sysctl.d",
"/lib/sysctl.d"

Any file in here with a .conf extension is valid fodder. But again... this all seems to be applied even when systemd is not pid 1. I guess I'm just leaving this here for anyone else who's interested.

The only other difference between the sysvinit and systemd setup is the numeric index given to eth0. Perhaps this is relevant... what other network devices are present? Does ethtool work on any of them?

diac · 2011-08-22 22:34:30

tomegun wrote:

I really don't get what is going on here, but I don't think udev is the one causing problems...

Yes, it seems.

falconindy wrote:

Any file in here with a .conf extension is valid fodder. But again... this all seems to be applied even when systemd is not pid 1. I guess I'm just leaving this here for anyone else who's interested.

Ok thank you, It's interesting for me !

falconindy wrote:

The only other difference between the sysvinit and systemd setup is the numeric index given to eth0. Perhaps this is relevant... what other network devices are present? Does ethtool work on any of them?

I can't see any other devices with ifconfig/ethtool/mii-tool, only wlan0 (on initV ad systemd) and eth0 (on initV only).

diac · 2011-08-23 12:13:40

I have check all the IFINDEX attributed by udev and the variation is due to the order of the interfaces eth0 and wlan0.
lo is always IFINDEX=1 but eth0 and wlan0 can be 2 or 3, it's depends.

Any more ideas ?

framas · 2011-08-24 05:09:42

falconindy wrote:

framas wrote:
I need some help,
systemd mounts my /media/ fstab entrys always with the noexec flag even though I put exec in the fstab entry
cat /etc/fstab| grep media
/dev/mapper/arch_lvm-data /media/work btrfs rw,suid,exec,auto,user,async,compress=lzo,ssd 0 0

mount | grep media
dev/mapper/arch_lvm-data on /media/work type btrfs (rw,noexec,nosuid,nodev,compress=lzo,ssd)
This also happens on my other pc (also arch with systemd) with a "normal" /media/ partition (non lvm).
Where can I change this behavior?
Not a systemd-specific behavior. You can change it in your fstab. 'user' implies 'noexec', and ordering prevails. 'user' should be first, if you're going to have it at all.
Aside, systemd creates a tmpfs on top of /media so that udisks can create directories and mount devices even when / is read only. If you have automounted partitions, I suggest mounting your static partitions in /mnt, not /media.
Fackamato wrote:
systemd does mounting? I thought it was an init system?
An init system reading /etc/fstab? Heresy. We definitely don't do this in /etc/rc.sysinit ........

Thanks falconidy. I've got it.

For me it seems to be first a systemd issue, because when I boot with sysinit the partition is mounted as expected.
Strange thing as you mentioned that ordering prevails.

diac · 2011-08-25 11:43:41

Hi,

I have find the guilty and systemd, dbus, udev are innocent.
In fact the guilty is laptop-mode demond ! I don't know why yet but I will look at it.

I free the topic cos I'm out subject.

Thanks to your help.

ganlu · 2011-08-27 08:46:16

I tried systemd(34-1), everything seems OK and faster, but gnome shell refuses to start and complaint lack of driver support. I check the system info from control center, it shows graphical driver is "llvmpipe" instead of Gallium 0.4 on ATI RV380 (I am using T43P), doesn't anyone know why? Thanks a lot.

herredgar · 2011-08-27 21:46:24

I got a problem with samba and systemd. nmbd fails to boot

# systemctl status nmbd.service
nmbd.service - Samba NetBIOS name server
          Loaded: loaded (/lib/systemd/system/nmbd.service; enabled)
          Active: failed since Sat, 27 Aug 2011 23:39:28 +0200; 1min 39s ago
         Process: 464 ExecStart=/usr/sbin/nmbd -F (code=exited, status=1/FAILURE                   )
          CGroup: name=systemd:/system/nmbd.service

doing a

systemctl restart samba.service

does work fine though

falconindy · 2011-08-27 21:49:12

ganlu wrote:

I tried systemd(34-1), everything seems OK and faster, but gnome shell refuses to start and complaint lack of driver support. I check the system info from control center, it shows graphical driver is "llvmpipe" instead of Gallium 0.4 on ATI RV380 (I am using T43P), doesn't anyone know why? Thanks a lot.

http://lists.freedesktop.org/archives/s … 03268.html

herredgar wrote:

I got a problem with samba and systemd. nmbd fails to boot

# systemctl status nmbd.service
nmbd.service - Samba NetBIOS name server
          Loaded: loaded (/lib/systemd/system/nmbd.service; enabled)
          Active: failed since Sat, 27 Aug 2011 23:39:28 +0200; 1min 39s ago
         Process: 464 ExecStart=/usr/sbin/nmbd -F (code=exited, status=1/FAILURE                   )
          CGroup: name=systemd:/system/nmbd.service

doing a

systemctl restart samba.service

does work fine though

You appear to be looking at 2 separate daemons here... samba.service likely refers to invocation of the legacy /etc/rc.d/samba initscript.

Last edited by falconindy (2011-08-27 22:00:11)

ganlu · 2011-08-28 00:46:52

Thanks.

dptkby · 2011-08-29 20:42:57

falconindy wrote:

http://lists.freedesktop.org/archives/s … 03268.html

Do you happen to know whether something similar applies to LightDM? For GDM, you can "fix" (I guess this is something of a workaround) the problem by editing /etc/pam.d/gdm, but there's not even a file for LightDM.

Edit: Oh, and did anything come of this? If I'm not mistaken, you didn't change it in the Arch package yet?

falconindy wrote:

Uggh...this is broken behavior upstream
https://bugzilla.redhat.com/show_bug.cgi?id=663900
Unsetting the locale environment vars is a filthy hack to avoid issues with indian characters which the terminal cannot display. I'm inclined to fix this in our package since I suspect that there's a larger ratio of archers logging in at a getty than there is for say, fedora.

Last edited by dptkby (2011-08-29 21:00:42)

sidneyk · 2011-09-02 15:45:03

I've looked high and low and tried many different things with no luck. I'm using Gnome3 with gdm and plymouth with systemd (and the plymouth specific pkgs for systemd and gdm that someone else here posted about). Does anyone know a simple way to get rid of the console output that happens on reboot, between gdm quitting and plymouth starting? The plymouth pkg. for gdm-plymouth moves the active tty to tty1 instead of tty7. It is that output that seems to have gotten much longer after recent updates that just says it shutting this and that down. I don't really need to see it unless there is some problem and even then I could get it from a log I suppose. Thanks in advance and apologies if this has been covered somewhere that I missed.

PIMPinator · 2011-09-02 17:24:39

Hey guys, just bought a new laptop so time for a new install. Should I be looking at systemd for an everyday system or is it still fairly buggy and unsupported?

Zom · 2011-09-02 18:21:39

PIMPinator wrote:

Hey guys, just bought a new laptop so time for a new install. Should I be looking at systemd for an everyday system or is it still fairly buggy and unsupported?

Try it for yourself, it's not like you can't go back to regular init if it craps out.

(well, unless you remove those packages yourself, but systemd doesn't conflict with it).

I think it works quite nice for everyday use, and I've been using it for a couple of months at least.

PIMPinator · 2011-09-02 19:55:15

Alright, looks like I'll have to give it a go.

MisterAnderson · 2011-09-04 16:17:42

I installed, found it a bit pointless for me for now and I randomly had no sound, so I uninstalled, and now I can't shutdown or reboot? Using the commands fails to do anything other than print the usual messages, and doing so from with xfce (launched with startx) simply kills X then leaves me at tty1 still logged in.

Last edited by MisterAnderson (2011-09-04 16:23:11)

gargamel · 2011-09-05 13:14:22

I had to add ExecStartPost=/bin/touch /run/daemons/rpcbind and ExecStopPost=/bin/rm /run/daemons/rpcbind to rpcbind.service to make nfs-common.service work.
this is due to nfs-common.service start rpcbind.service and then run /etc/rp.c/nfs-common start (expecting /etc/rp.c/rpcbind to start and make that file)

For dnsmasq i had to
change to <policy user="dnsmasq"> in /etc/dbus-1/system.d/dnsmasq.conf
adapt dnsmasq.conf user=dnsmasq
adduser dnsmasq
else sudo /etc/rc.d/dnsmasq start is ok but sudo systemctl start dnsmasq.service fail after 2 min

gargamel · 2011-09-05 14:08:25

systemd[1]: dnsmasq.service operation timed out. Terminating.
It was not the good solution.

falconindy · 2011-09-05 14:18:10

You need to enable dbus in dnsmasq's config (/etc/dnsmasq.conf). There's no need to fart around with dbus config. This will hopefully be solved with some tweaks to the upstream unit file... In the meantime, this is what I've been using:

[Unit]
Description=A lightweight DHCP and caching DNS server

[Service]
Type=dbus
BusName=uk.org.thekelleys.dnsmasq
ExecStartPre=/usr/sbin/dnsmasq --test
ExecStart=/usr/sbin/dnsmasq -k --enable-dbus --user=dnsmasq --pid-file
ExecReload=/bin/kill -HUP $MAINPID

[Install]
WantedBy=multi-user.target
Alias=dbus-uk.org.thekelleys.dnsmasq

You'll need to create the dnsmasq user yourself.

Last edited by falconindy (2011-09-05 17:19:48)

MisterAnderson · 2011-09-06 11:14:49

I fixed my issue. I had to reinstall systemd and reboot without it in the kernel line to uninstall.

EDIT: Although I'd like to move to systemd the Wiki page doesn't give a clear explanation of what it is or does, and a poor explanation of how to install and what to expect.

Last edited by MisterAnderson (2011-09-06 12:34:13)

falconindy · 2011-09-06 13:48:36

The wiki page stinks. I hate that it's a giant FAQ of nothingness, but I haven't really made any progress in fixing that (maybe someone else will?). imo it should cover, at least:

* requirements (kernel, in particular)
* base package installation (explanation of the 3 packages in community, with mention of -git equivalents)
* native configuration files and directories, aka how to get rid of rc.conf
* daemon => service migration
* a brief mention of the various tools packaged -- systemd-analyze, systemctl, systemd-loginctl, systemd-cgls...
* imo a section on automounting via /etc/fstab is warranted, as it isn't really documented in systemd

We've got some of this already. A lot of it is really just restructuring. We shouldn't re-explain everything that can be gleaned from say, Lennart's blog posts. His administrator and developer series is more than sufficient to explain the whole what/why/how of systemd in combination with a link to the kernel doc on Cgroups. The man pages, for the most part, are extremely thorough, but worth reading. I almost always find what I'm looking for.

MisterAnderson · 2011-09-06 19:04:49

I'd do some of the wiki page myself but since I have no idea what the point of systemd is I can't really help. In this instance, at the very least a brief explanation of what it is and why to install it would be great. Most other pages include this kind of thing.

falconindy · 2011-09-06 20:30:05

Fedora's done a nice job of answering most of your questions. Though it's a bit dated, it's still applicable.

http://fedoraproject.org/wiki/Systemd

Last edited by falconindy (2011-09-06 20:30:23)

gee · 2011-09-07 02:56:03

Hey,

Booting is ok in terms of speed, but shutdown can be super slow.
It's been like that only for a few weeks so I am not sure what happened...
is there anything like systemd-analyze blame but for restarting?
I am already using rsyslog...
Thanks!

MisterAnderson · 2011-09-08 12:38:23

Thanks for the link falconindy. For anyone wondering about my earlier problem of not shutting down is that I was using halt and not poweroff.

flamelab · 2011-09-08 12:45:48

Question: I start KDM through inittab. Is it read by systemd or isn't it? If it's not, I use KDM in DAEMONS array and it is parsed by systemd ?

Arch Linux

#926 2011-08-22 22:00:22

Re: systemd: Yet Another Init Replacement

#927 2011-08-22 22:34:30

Re: systemd: Yet Another Init Replacement

#928 2011-08-23 12:13:40

Re: systemd: Yet Another Init Replacement

#929 2011-08-24 05:09:42

Re: systemd: Yet Another Init Replacement

#930 2011-08-25 11:43:41

Re: systemd: Yet Another Init Replacement

#931 2011-08-27 08:46:16

Re: systemd: Yet Another Init Replacement

#932 2011-08-27 21:46:24

Re: systemd: Yet Another Init Replacement

#933 2011-08-27 21:49:12

Re: systemd: Yet Another Init Replacement

#934 2011-08-28 00:46:52

Re: systemd: Yet Another Init Replacement

#935 2011-08-29 20:42:57

Re: systemd: Yet Another Init Replacement

#936 2011-09-02 15:45:03

Re: systemd: Yet Another Init Replacement

#937 2011-09-02 17:24:39

Re: systemd: Yet Another Init Replacement

#938 2011-09-02 18:21:39

Re: systemd: Yet Another Init Replacement

#939 2011-09-02 19:55:15

Re: systemd: Yet Another Init Replacement

#940 2011-09-04 16:17:42

Re: systemd: Yet Another Init Replacement

#941 2011-09-05 13:14:22

Re: systemd: Yet Another Init Replacement

#942 2011-09-05 14:08:25

Re: systemd: Yet Another Init Replacement

#943 2011-09-05 14:18:10

Re: systemd: Yet Another Init Replacement

#944 2011-09-06 11:14:49

Re: systemd: Yet Another Init Replacement

#945 2011-09-06 13:48:36

Re: systemd: Yet Another Init Replacement

#946 2011-09-06 19:04:49

Re: systemd: Yet Another Init Replacement

#947 2011-09-06 20:30:05

Re: systemd: Yet Another Init Replacement

#948 2011-09-07 02:56:03

Re: systemd: Yet Another Init Replacement

#949 2011-09-08 12:38:23

Re: systemd: Yet Another Init Replacement

#950 2011-09-08 12:45:48

Re: systemd: Yet Another Init Replacement

Board footer