no shutdown at all in fedora 24 Again!

[François Patte]

Bonjour,

I recall my last posts about this problem (complete shutdown impossible
(fedora 24)).

In his hanswer, Chris Murphy suggested to use these tests :

https://freedesktop.org/wiki/Software/systemd/Debugging/#index2h1

First test: sync && reboot -f shows that it is not a kernel bug.

I went to the page listing the submitted bugs for systemd and I found
this answer to a similar problem:

<quote from Kay Sievers>
Systemd does not support "fake block devices". This needs to be provided by
the raid tools themselves and hook up into systemd.

Most of the things are done today by (mostly weird) shell scrips provided
by the distributions. Systemd has currently no plans to support MD or LVM
natively.

We provide minimal dis-assembly support for device-mapper volumes, something
similar could probably be done for MD, but someone needs to do it, and it
would need to be reasonable simple and not involve running daemons,
calling-out
to command line tools or linking against uncommon libraries.
</quote>

So, it seems that the bug is from Fedora and some "*mostly weird shell
scrips provided
by the distributions*".

Where do I report my problem (with tons of log provided by journalctl,
logwatch, and some other tests...)

It could be kind from fedora to say that Raid is not fully  supported by
systemd and to provide an alternative if people want to have a RAID
system (with encryption).

For how many years now systemd has been adopted by Fedora?

Thank you for attention.



-- 
François Patte
UFR de mathématiques et informatique
Laboratoire CNRS MAP5, UMR 8145
Université Paris Descartes
45, rue des Saints Pères
F-75270 Paris Cedex 06
Tél. [phone]
http://www.math-info.univ-paris5.fr/~patte

[Chris Murphy]

I think you're going to have to put logs up somewhere and
configuration information somewhere to get an answer. I can tell you
that mdadm raid and lvm being used do not themselves result in an
inability to shutdown Fedora. Fedora by default uses LVM, and mdadm
raid0/1/5 tests are done on pre-release versions of Fedora; if any of
that resulted in an inability to restart or shutdown, it'd likely be a
release blocking bug. So the mere fact systemd itself does not
directly support md raid or lvm doesn't really translate into a cause
of your problem.


Chris Murphy

[Ralf Corsepius]

On 01/22/2017 02:26 AM, Chris Murphy wrote:
>  I think you're going to have to put logs up somewhere and
>  configuration information somewhere to get an answer. I can tell you
>  that mdadm raid and lvm being used do not themselves result in an
>  inability to shutdown Fedora. Fedora by default uses LVM, and mdadm
>  raid0/1/5 tests are done on pre-release versions of Fedora; if any of
>  that resulted in an inability to restart or shutdown, it'd likely be a
>  release blocking bug. So the mere fact systemd itself does not
>  directly support md raid or lvm doesn't really translate into a cause
>  of your problem. 
I am also facing these shutdown probs and have come to the conclusion, 
that there is "no single cause", but that they are the result of a 
combination of several causes, probably related to systemd, 
NetworkManager, SELinux and other, networking related packages.

Isolating them seems to be difficult, because almost each update changes 
and breaks something else.

E.g. right now, rpcbind/nfs/yp/nis's integration into systemd seems to 
be "mash" in fc25 - It has "bricked" one of my machines, today!

Ralf

[François Patte]

Le 22/01/2017 02:26, Chris Murphy a écrit :
>  I think you're going to have to put logs up somewhere and
>  configuration information somewhere to get an answer. I can tell you
>  that mdadm raid and lvm being used do not themselves result in an
>  inability to shutdown Fedora. Fedora by default uses LVM, and mdadm
>  raid0/1/5 tests are done on pre-release versions of Fedora; if any of
>  that resulted in an inability to restart or shutdown, it'd likely be a
>  release blocking bug. So the mere fact systemd itself does not
>  directly support md raid or lvm doesn't really translate into a cause
>  of your problem. 
I submitted a bug report to Redhat-bugzilla with a 3750 lines file
attached (result of lsblk command, dmesg result, and logwatch report
concerning systemd log after a shutdown failure.

But I am a little bit pessimistic if you consider that (for instance)
since f22, you can see that journalctl or watchdog are not stopped
before unmounting /  /var and /usr and you can see while you shutdown
your system (f23, f24) that unmounting these partitions systematically
fails.... This kind of small bugs remain after 4/5 years....

I am wondering why it is claimed that systemd has been introduced to
increase fastness: I could see that it is not the case at boot time (10s
with systemv while it is now about 30s when no "start job ......"
appears during the boot sequence!) the last update I made on a f23
system blocks change of user (I have to beboot) and shutdown is very
long (no raid on this system), and now, endless shutdown on my f24 system!

You say that fedora uses LVM by default, but on my f24 system (raid)
there are no LVM at all: pvscan, lvscan etc. return nothing while some
lvm services are enabled and running (lvm2-lvmetad.socket,
lvm2-monitor and lvm2-lvmpolld) what for?

Hope that someone could help!


-- 
François Patte
UFR de mathématiques et informatique
Laboratoire CNRS MAP5, UMR 8145
Université Paris Descartes
45, rue des Saints Pères
F-75270 Paris Cedex 06
Tél. [phone]
http://www.math-info.univ-paris5.fr/~patte