1 |
Hello, folks. I sent this to the list 2 weeks ago, and I'm not sure it came |
2 |
through. |
3 |
I'm resending it now. I'd be greatly pleased if someone could help me on |
4 |
this issue. |
5 |
|
6 |
|
7 |
We have a number of old machines hanging diskless on a fairly nice LTSP box. |
8 |
So, this is a good server with a bunch of 6-7 machines using X remotely. |
9 |
No problem at all, everything working fine... |
10 |
... that is, until one of them hangs. |
11 |
Certain old machines tend to hang a lot, and when they hang, their processes |
12 |
running |
13 |
on the server don't die, or at least don't die immediately. |
14 |
When they are rebooted and log on again, certain applications won't run, |
15 |
either |
16 |
complaining about lock files (OO.org), about another running instance of |
17 |
itself |
18 |
(firefox), or mibehaving because of existing files in /tmp (gnome with orbit |
19 |
files). |
20 |
There are other applications behaving like this. |
21 |
Anyone here using LTSP ever done any workaround to this? I thought about |
22 |
making |
23 |
a wrapper script to gdm so that when someone logs on, before starting |
24 |
anything at |
25 |
all (window manager et all), it would kill every running process on the |
26 |
server except |
27 |
itself, and only then pass control to the window manager. |
28 |
This is an ugly hack, I know, but I don't see a straightforward solution |
29 |
(apart from |
30 |
buying loads of new expensive machines). It seems to be a problem between X |
31 |
client |
32 |
and X server. In my opinion, the client windows should die if the |
33 |
corresponding X server |
34 |
is not there anymore, but that's not what happens. |
35 |
|
36 |
-- |
37 |
Bruno Lustosa, aka Lofofora | Email: bruno@×××××××.net |
38 |
Network Administrator/Web Developper | ICQ: 1406477 |
39 |
Rio de Janeiro - Brazil | |