lubuntu-desktop team mailing list archive

[Jonathan Marsden <jmarsden@xxxxxxxxxxx>]

On 03/15/2011 06:16 PM, Jared Norris wrote:

> I am experiencing some really odd behaviour lately and I wanted some 
> ideas on how to troubleshoot it further. Every now and then,
> seemingly randomly, the mouse and keyboard will really start to lag,

How often is "every now and then"?  Once a day?  Once a month? :)  If it
is very infrequent, you will probably need to run monitoring tools all
the time, so you can "catch it in the act".

> ...The second test was to SSH into the machine and the input is
> severley lagged ...

If SSH shell response time is affected, then I would think you have
shown that it is not "just the keyboard and mouse", since you were using
a different keyboard and mouse on a different PC at that point.

In that case, either the issue is "keyboard, mouse and network" (which
would be fairly odd), or the machine is busy doing something else when
you are seeing the lag.  A really large disk i/o load can make Linux
boxes pretty unresponsive, especially for machines with just one PATA or
SATA hard drive.

> ...can anyone suggest how I can trouble shoot this problem further?

(1) Based on the above, I would suggest running system monitoring tools
to check what else the machine is doing when it is lagging your
keyboard/mouse input.  htop, iotop, jnettop will do for simple
text-based "what is doing a lot of work on my PC" checking. vmstat and
iostat might be worth a look too.

(2) If you are comfortable working at the command line, you could also
consider running the machine with no GUI (no X) -- no LXDE stuff at all,
just plain old text mode consoles) for a while.  If the issue goes away,
you would then suspect that whatever is causing the problem is X related
in some way.  With enough time, you could then start X and run just an
xterm or LXterminal window and see if that lags or not... and so on...
slowly building back up towards a full LXDE GUI.  Knowing what was added
that started the problem up again would be a very big clue.

(3) If you have another spare test machine, try setting it up as close
to identical to the first one as you can, documenting how to do that
step by step.  Then, see if you can reproduce the problem on the second
machine too.  If you can reproduce it there, you now have accurate and
tested step-by-step "how to reproduce" information for the bug report,
which could be very handy for others trying to duplicate and track down
the issue :)

Realistically, this kind of "weird stuff happens occasionally" issue is
going to be hard to track down.  Hopefully the above suggestions will
help, if you decide you have the patience to really work on doing that.

Jonathan