Bug#808151: systemd: failed to start remount root and kernel file system

[Michael Biebl]

Here is a more complete log excerpt for v228 (full log attached)

> Dez 20 01:27:42 debian systemd[1]: -.mount: Changed dead -> mounted
> Dez 20 01:27:42 debian systemd[1]: usr.mount: Changed dead -> mounted
> Dez 20 01:27:42 debian systemd[1]: usr.mount: Unit is bound to inactive unit dev-sda5.device. Stopping, too.
> Dez 20 01:27:42 debian systemd[1]: usr.mount: Trying to enqueue job usr.mount/stop/fail
> Dez 20 01:27:42 debian systemd[1]: usr.mount: Installed new job usr.mount/stop as 1
> Dez 20 01:27:42 debian systemd[1]: usr.mount: Enqueued job usr.mount/stop as 1
> Dez 20 01:27:42 debian systemd[1]: dev-sda1.device: Changed dead -> tentative
> Dez 20 01:27:42 debian systemd[1]: -.slice changed dead -> active
> Dez 20 01:27:42 debian systemd[1]: dev-sda5.device: Changed dead -> tentative
> Dez 20 01:27:42 debian systemd[1]: init.scope changed dead -> running
> Dez 20 01:27:42 debian systemd[1]: Activating default unit: default.target
> Dez 20 01:27:42 debian systemd[1]: graphical.target: Trying to enqueue job graphical.target/start/isolate
> Dez 20 01:27:42 debian systemd[1]: display-manager.service: Cannot add dependency job, ignoring: Unit display-manager.service failed to load: No such file or directory.
> Dez 20 01:27:42 debian systemd[1]: usr.mount: Job usr.mount/stop finished, result=canceled

So in my case the stop request is cancelled, probably because the state
has changed from dead to tentative quickly enough. So this looks like a
race indeed.

I also tested v227 and it shows exactly the same behaviour. So this
issue has been lurking for a while (see attached journal) it seems.
My guess is that it goes way back until [1] or even [2].

So the best guess is that the timing in v228 changed a little (some code
paths became faster). This would confirm Frank's findings that enabling
verbose output (which slows down boot a bit) made it less likely to hit.

Martin has been working/debugging the tentative stuff in the past, so
maybe he has some insights here.

We should probably also involve upstream at some point.


[1] https://github.com/systemd/systemd/commit/f620094
[2] https://github.com/systemd/systemd/commit/628c89c

-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: journal-228.xz
Type: application/x-xz
Size: 51760 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/pkg-systemd-maintainers/attachments/20151220/34032b8b/attachment.xz>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: journal-227.xz
Type: application/x-xz
Size: 34424 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/pkg-systemd-maintainers/attachments/20151220/34032b8b/attachment-0001.xz>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://alioth-lists.debian.net/pipermail/pkg-systemd-maintainers/attachments/20151220/34032b8b/attachment-0002.sig>