1 |
On Jan 15, 2013 7:59 PM, "Alan McKinnon" <alan.mckinnon@×××××.com> wrote: |
2 |
> |
3 |
> On Tue, 15 Jan 2013 11:09:56 +0000 |
4 |
> Neil Bothwick <neil@××××××××××.uk> wrote: |
5 |
> |
6 |
> > On Tue, 15 Jan 2013 11:57:21 +0200, Alan McKinnon wrote: |
7 |
> > |
8 |
> > > On the rare occasion when I reboot or shut this laptop down, it |
9 |
> > > continually and consistently gets stuck on one of the final steps, |
10 |
> > > to umount /home |
11 |
> > |
12 |
> > If you logout as your user(s) so only root is logged in, does lsof |
13 |
> > show any hits for /home? |
14 |
> |
15 |
> Only 1 hit - a background ssh process that sets up a bunch of tunnels |
16 |
> and port forwards so I can get into the corporate network for anywhere. |
17 |
> |
18 |
> But I was not able to make the problem re-appear in short reboot |
19 |
> cycles. So whatever is hanging the box is something that starts up in |
20 |
> the course of work, it doesn't appear to be there directly after a KDE |
21 |
> login. |
22 |
> |
23 |
> I might have to look at my work flow closely to find all those unusual |
24 |
> things I do in the course of work. I did manage to see the full |
25 |
> shutdown output on the screen with these tests though - it |
26 |
> umounts /home then umounts /boot, but the second umount message is |
27 |
> never displayed. Seeing as it's just a regular sdb1 partition using |
28 |
> ext2 I don;t really think it's getting stuck as the next step starts |
29 |
> up. It's more likely to have something to do with /home |
30 |
> |
31 |
> |
32 |
> > |
33 |
> > I had a similar problem with my MythTV backend failing to |
34 |
> > unmount /var. It turned out that mythbackend was failing to shutdown |
35 |
> > but openrc carried on trying to shutdown. If I make sure mythbackend |
36 |
> > really is stopped, the reboot proceeds normally, which is much better |
37 |
> > since I had to go into the loft to reboot the box manually, |
38 |
> > |
39 |
> > |
40 |
> |
41 |
> |
42 |
> |
43 |
> -- |
44 |
> Alan McKinnon |
45 |
> alan.mckinnon@×××××.com |
46 |
> |
47 |
> |
48 |
|
49 |
A bit roundabout, but you can also try making a 'pseudo-service'. Make it |
50 |
'depend' on a late-stage service so it starts last, and shuts down early. |
51 |
The stop() part of the pseudo-service should perform an lsof >> a file (in |
52 |
a directory still available during the last throes of OpenRC like, say, |
53 |
/etc). |
54 |
|
55 |
I hope I'm making sense... |
56 |
|
57 |
Rgds, |