Comment 113 for bug 1073433

Steve, thanks for the info.
As I've mentioned earlier, I'm using autofs. I think that such names like 'tmp/autoXXXXX' are autofs temporary mount points. Also I have /tmp mounted as tmpfs.
So my guess is that tmpfs is getting unmounted BEFORE autofs daemon have been terminated (unmounted it's mountpoints). So that autofs mount points can't be unmounted (because they were left on old /tmp with tmpfs).

Now starting and stopping of services in Ubuntu not always works as it is expected. Earlier when services got started from /etc/init.d it was slow but predictable. Services have been launched one by one.
What is really going on now?.. Looks like no one really knows. Sometimes it looks like that all this stuff is launched asynchronously (simultaneously) and in background. Some kind of unpredictable hell, full of 'race conditions'. And if you don't see it that doesn't mean that this is not exist.
For me it's like playing with fire. Today you have won... But tomorrow you can burn.
For an example, I've seen a lot of times, when I do 'restart service_name' command I'm getting almost immediately output to console that service have been restarted. But meanwhile, when I've tried to connect to that service it wasn't ready! It is still being launched/restarted!
So why do I get output to the console, that the service have been restarted when in fact it still DOESN'T?

I'm just suspecting that the problem with 'busy' can be even deeper than I thought...