Hi,
I have a home server which runs F17 and for a long time now it just doesn't reboot or shutdown cleanly. I've had the same issue since it was running F16 and possibly F15. When the server shuts down, I get spurious messages about some services failing or a dependency was not met for the shutdown target, or sometimes a black screen with a line which just says "Adding Swap target" (really, on a shutdown?). The output seems to vary every time, sometimes wildy different but the end result is often the same, having to hit the reset button. How do I about starting to diagnose these problems with systemd? Given that the system is left in a state where I can't do anything with it. The usual places such as /var/log/messages doesn't turn up anything useful. Maybe because syslog has already been stopped by then.
On 27.10.2012, Ian Chapman wrote:
I have a home server which runs F17 and for a long time now it just doesn't reboot or shutdown cleanly. I've had the same issue since it was running F16 and possibly F15. When the server shuts down, I get spurious messages about some services failing or a dependency was not met for the shutdown target, or sometimes a black screen with a line which just says "Adding Swap target" (really, on a shutdown?).
I have encountered similar problems with systemd, see e.g. here for the latest one which is present a long time:
https://bugzilla.redhat.com/show_bug.cgi?id=857406
Neither Fedora nor sysyemd-people really care, so expect to be on your own.
How do I about starting to diagnose these problems with systemd?
http://freedesktop.org/wiki/Software/systemd/Debugging#Diagnosing_Shutdown_P...
Given that the system is left in a state where I can't do anything with it. The usual places such as /var/log/messages doesn't turn up anything useful. Maybe because syslog has already been stopped by then.
Do the diagnostics from the link above and open a bugreport. Maybe you are more lucky than I have been. In the meantime, I switched all my machines to Archlinux, there's just one Fedora machine left (which will go soon). Most of the showstoppers were systemd ones, where the developers/maintainers didn't care.
"Didn't care" translates in my case *not* to "they do not provide the help and fixes I'm expecting" (it's free software, after all). It translates to "I don't even get an answer"", despite that I offered full support to debug the case.
On Sat, Oct 27, 2012 at 09:36:20PM +0200, Heinz Diehl wrote:
I have encountered similar problems with systemd, see e.g. here for the latest one which is present a long time: https://bugzilla.redhat.com/show_bug.cgi?id=857406 Neither Fedora nor sysyemd-people really care, so expect to be on your own.
This isn't true. I'm a Fedora person, and I care very much. And, I know that I'm one of many. It's a fact that some bug reports and offers of help are left hanging, and obviously that's not good. We need to do better in this area.
On 28/10/12 03:36, Heinz Diehl wrote:
I have encountered similar problems with systemd, see e.g. here for the latest one which is present a long time:
https://bugzilla.redhat.com/show_bug.cgi?id=857406
Neither Fedora nor sysyemd-people really care, so expect to be on your own.
Yeah I've just lived with it so far, every now and then seeing if I can fix it. Luckily as it's a server I don't need to take it down all that often but I loathe having to do it. Unfortunately raising bugs is a bit hit and miss.
How do I about starting to diagnose these problems with systemd?
http://freedesktop.org/wiki/Software/systemd/Debugging#Diagnosing_Shutdown_P...
Thanks for that link, there's a few things in there worth looking at. I'll see if I can get any additional information with it. I almost wish there was a fay to force systemd to start & stop services serialized as half the battle seems to be not knowing what systemd is actually doing at any point in time because of its parallel nature.
On 27.10.2012, Matthew Miller wrote:
Neither Fedora nor sysyemd-people really care, so expect to be on your own.
This isn't true. I'm a Fedora person, and I care very much. And, I know that I'm one of many.
I know. I was unreflective and wrote this statement as a generalization, which it shouldn't have been. Sorry for that!
It's a fact that some bug reports and offers of help are left hanging, and obviously that's not good. We need to do better in this area.
It's quite clear to me that there are lots of problems, much more than people can (re)solve. My point here is that people actually have seen the reports and requested more information, which implies that they work on it. And then, all is dead. A simple "we have no time" or "we can't help you" would be suffice in such cases.
-
Heinz Diehl -
Ian Chapman -
Matthew Miller