You are not logged in.
I'm trying to clean up my systemd workings as I've mentioned in other posts. Anyway, I did this:
systemctl list-units --all |grep error
auditd.service error inactive dead auditd.service
display-manager.service error inactive dead display-manager.service
plymouth-quit-wait.service error inactive dead plymouth-quit-wait.service
plymouth-start.service error inactive dead plymouth-start.service
syslog.service error inactive dead syslog.service
I decided to look at plymouth. "Pacman -Qs plymouth" yields nothing, which is no surprise as it is not the kind of thing that appeals to me, so I wouldn't have installed it. Yet since Sept 5 my system log has this:
systemd[1]: Failed to load configuration for plymouth-quit-wait.service: No such file or directory
...and also for the start service.
I found out that (for example), /etc/systemd/system/getty.target.wants/getty@tty1.service has a line in it:
After=systemd-user-sessions.service plymouth-quit-wait.service
So it appears some unit files call plymouth even though plymouth was never installed, leading to those errors in the system log and in systemctl.
Is this normal? Are we supposed to put up with these sorts of errors in our logs or is there a way to clean this up?
Last edited by PaulBx1 (2013-01-23 15:53:28)
Offline
I don't see anything about plymouth in my systemd journal, but mine only goes back to early Dec. Are you sure those errors are still occuring? Maybe it was just an old bug that's been fixed.
Also, as I understand it, services on an After= line are not "called" (ie. systemd does not attempt to start them).
Last edited by alphaniner (2013-01-21 20:42:44)
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
You are right, the system log references to plymouth go only to Oct 25, sorry I assumed they continued.
However systemctl still says it's an error. Perhaps you are right about the "After" line, but I'm guessing there is some reference or call to plymouth either in the /etc/systemd or in /usr/lib/systemd files, that causes systemctl to throw an error. To me this is... sub-optimal. Errors should mean something. It gets hard to separate the wheat from the chaff otherwise.
What do you get with "systemctl list-units --all |grep error"?
Offline
I found this thread on the same subject:
https://bbs.archlinux.org/viewtopic.php?id=146295
I'm trying to digest it...
Offline
What do you get with "systemctl list-units --all |grep error"?
The same as you.
Out of curiosity, I added "i-heart-fan.service" to the After= line of my getty@.serivce file. Rebooted and:
$ systemctl list-units --all |grep error
...
i-heart-fan.service error inactive dead i-heart-fan.service
...
So it's possible that the only reason the others are there is because they are mentioned in After= (or maybe Before= etc.) lines as well.
Edit: I looked over that thread. Masking works, but the way it's described in the manual it seems a rather extreme method for silencing an error message.
Last edited by alphaniner (2013-01-21 21:50:07)
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
I wonder what other kinds of things can generate an "error"? Or is this the only thing?
It really is an error, referencing something that ain't there. But the remedies (that I can imagine anyway) don't seem very good.
Maybe this kind of "error" should be called "warning" or "missing" instead, so people won't get concerned over nothing.
Last edited by PaulBx1 (2013-01-21 22:15:00)
Offline
IMO, a reference in the After= line to a service that doesn't exist shouldn't even be considered an error.
And I wonder why a stock service would reference something like plymouth in the first place.
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
I wonder why a stock service would reference something like plymouth in the first place.
Because that stock service is written in the context of a distro where plymouth is also a stock service and installed by default - isn't it?
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
I thought plymouth was an Ubuntu thing, but a cursory search seems to suggest it's from Red Hat. Guess it makes sense, as they employ Poettering.
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
$ systemctl list-units --all | grep error
auditd.service error inactive dead auditd.service
plymouth-quit-wait.service error inactive dead plymouth-quit-wait.service
plymouth-start.service error inactive dead plymouth-start.service
I'm curious about what the "--all" refers to. I'd always figured it meant something like "all units requested" or "all units enabled/started" or something similar. But it must mean something else unless just being listed in an "After=" counts as being enabled/started/requested in the relevant sense.
It is confusing, too, because "inactive dead" suggests to me that it was once "active alive" and not that execution (?) failed...
CLI Paste | How To Ask Questions
Arch Linux | x86_64 | GPT | EFI boot | refind | stub loader | systemd | LVM2 on LUKS
Lenovo x270 | Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz | Intel Wireless 8265/8275 | US keyboard w/ Euro | 512G NVMe INTEL SSDPEKKF512G7L
Offline
"I was dead for millions of years before I was born and it never inconvenienced me a bit." — Mark Twain.
Aparently unborn systemd units may be a bit more of an inconvenience.
"UNIX is simple and coherent..." - Dennis Ritchie, "GNU's Not UNIX" - Richard Stallman
Offline
But at least he was going to be born. Besides, I'm prepared to grant Twain a degree of latitude I don't think should be extended to systemd units.
Lucretius thought the time before birth equivalent to the time after death but failed, like Twain, to explain how to apply this to systemd units.
Last edited by cfr (2013-01-22 02:37:48)
CLI Paste | How To Ask Questions
Arch Linux | x86_64 | GPT | EFI boot | refind | stub loader | systemd | LVM2 on LUKS
Lenovo x270 | Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz | Intel Wireless 8265/8275 | US keyboard w/ Euro | 512G NVMe INTEL SSDPEKKF512G7L
Offline
Does it make sense to just create dummy plymouth files there (which reference nothing else), with a note in them that it is to avoid throwing errors? Then if I ever do install plymouth the real unit file will simply overwrite the dummy.
In fact one could create a file called "dummy.service" and explain this business in a comment in it, telling the user to copy dummy.service to any other filename needed to eliminate the errors.
Last edited by PaulBx1 (2013-01-22 15:05:40)
Offline
Does it make sense to just create dummy plymouth files there (which reference nothing else), with a note in them that it is to avoid throwing errors? Then if I ever do install plymouth the real unit file will simply overwrite the dummy.
In fact one could create a file called "dummy.service" and explain this business in a comment in it, telling the user to copy dummy.service to any other filename needed to eliminate the errors.
I agree that "error" is kind of misleading, but come on, you only see these if you use "--all" which you usually don't even have a reason to use.
Creating dummy files just hides the "errors". It doesn't help you in any way and might even cause problems later on when you have all forgotten that you even created these files.
Seriously, just don't use "--all" and be done with it. If it really bothers you that much, talk to upstream...
just my 2 cents...
Offline
It doesn't bother me any more, now that I know what is going on. If you guys don't mind answering the same questions over and over, I don't mind that either.
I'm thinking this would be distro- and release-specific, since each distro (I think) supplies as default a different set of unit files, which would mean a different set of dummies. So I don't think it's an upstream thing (other than perhaps, explaining in the man page how "--all" can generate bogus errors, or that "--all" should only be used by experts!).
Last edited by PaulBx1 (2013-01-22 18:20:11)
Offline
I'm thinking this would be distro- and release-specific, since each distro (I think) supplies as default a different set of unit files
As far as sets, maybe. But the unit files themselves are written upstream and designed to be used without modification. That's one of the merits of systemd in contrast to SysV, for which it was necessary to maintain distro-specific init scripts. That's my understanding at least.
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
But the unit files themselves are written upstream and designed to be used without modification.
Who's talking about modification? I sure wasn't. I was talking about addition of dummy files.
The files chosen to be included in the set, even if each of them was written upstream, determines the number of errors one is going to see. The problem is that the chain of file references is broken rather than ended cleanly. That's the "error".
Once the default (minimal) set gets cleaned up this way, it should stay clean as new programs are added, because pacman takes care of dependencies. However I don't know if pacman looks at unit files to ensure all the correct dependencies are there too. Maybe it should. If it did, it would stay clean of this error no matter how many programs (and thus, unit files) were installed.
Is the default, minimal set determined by the distro, or do all distros simply follow upstream choices for that? If the latter, then it is an upstream problem. But that seems doubtful to me.
Last edited by PaulBx1 (2013-01-22 23:18:13)
Offline
Slightly relevant to OP: Pottering answers criticism of systemd, including the inclusion of plymouth related stuff.
Offline
Heh, well I wasn't intending to criticize systemd by any means. I brought up plymouth only as an example. I don't even mind that some plymouth unit files are part of the package; the more, the merrier. I just wanted to see what these errors were about, and suggest a way to eliminate them.
I have to say in general I do not like the current situation where we tolerate errors and warnings in all our logs. I had an old car that had a "check engine" light on all the time. I knew what the error code was and there was a good reason for it; however the utility of the "check engine" light was lost because, always being on for that reason, it masked other codes that I might have wanted to know about. I know the task is futile, but now and then I make a go at trying to clear out error codes in my logs, with limited success.
Offline
Hi everyone.
I don't like error messages either so I edited the following files to prevent the "plymouth" error.
Just cut the "plymouth *** references or put a hash in front of any lines that are only relevant to "plymouth" .
console-getty.service
console-shell.service
rescue.service
serial-getty@.service
getty@.service
emergency.service
systemd-ask-password-console.path
systemd-ask-password-console.service
All working OK . Booting OK no errors. Wether it's right or wrong depends on the individual.
NS
Last edited by NoSavvy (2013-01-23 22:37:50)
Offline
I hope I'm misunderstanding, because it sounds like you're suggesting commenting out entire lines, ie
#After=systemd-user-sessions.service plymouth-quit-wait.service
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline
Sorry.!! I cut out or hashed out ONLY the Plymouth references.. ie. plymouth-quit-wait.service as in your post.
Offline
You (presumably) would not need to edit any standard Arch-supplied file if you just created a dummy plymouth file in the same directory. I think editing these files is the wrong way to go.
Offline
Slightly relevant to OP: Pottering answers criticism of systemd, including the inclusion of plymouth related stuff.
One thing I thought was interesting about this was the claim that (something like) plymouth is really necessary for a "modern" boot process where I take it that "modern" means one which uses parallelisation. But I didn't fully follow it. I assume the thought is that it is somehow necessary for a smooth/optimal boot but I'm not altogether sure why - I guess it has something to do with multiple parallel stuff having to be filtered into a single output on display etc. but I'm not altogether clear how plymouth is supposed to help with that. As I understand it, plymouth is a kind of splash screen. If that's right, doesn't it hide that output rather than managing it? And why is that better than having the output even if it is a bit mixed up? And doesn't the kernel deal with parallel processes all the time anyway? Why is boot so different?
EDIT: the paragraph I'm thinking of is actually another follow-up in that thread at http://lists.debian.org/debian-devel/20 … 00349.html (not Pottering's).
Last edited by cfr (2013-01-23 23:45:46)
CLI Paste | How To Ask Questions
Arch Linux | x86_64 | GPT | EFI boot | refind | stub loader | systemd | LVM2 on LUKS
Lenovo x270 | Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz | Intel Wireless 8265/8275 | US keyboard w/ Euro | 512G NVMe INTEL SSDPEKKF512G7L
Offline
I think their definition of "modern" also entails a "user experience" palatable to the average Joe.
Also, I think it's incorrect to say that plymouth "hides ... output rather than managing it." The output can be brought forward with a single keystroke, and (at least according to an old phoronix article) the is logged exactly as it is/would_be displayed during boot.
But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner
Offline