[lug] load average constant at 2

[Ryan Kirkpatrick]

On Thu, 14 Sep 2000, Jason Bowen wrote:

> I pin-pointed what it was I think.  When httpd is starting there is a
> message about trying to mount on /netfs and that it will keep trying.  I
> didn't even have netfs starting up.  I killed amd and the load dropped to
> normal levels, start it back up and it went back up.  I'm going to reboot
> with netfs and amd starting at boot time to see if the problem persists.
> Thanks for all the help again.

	I remember reading some where that for every blocked process (has
a state of D in ps output), the load average is incremented by one. I have
seen this happen when a tape drive command jams, i.e. 'mt seek 12345',
tape drive gets lost and gives up, and mt is now stuck in the process
table with a 'D'. You can't kill it, even as root with kill -9, one has to
reboot. At the same time, my load average went straight to one with hardly
any CPU time being used.
	Attempting to perform mounts of network filesystems could result
in these blocked processes as well, and hence the source of your problems.
TTYL.

	PS. The tape drive in question above is a piece #$%@#% IDE TR-4
drive which I don't use any more. :)

---------------------------------------------------------------------------
|   "For to me to live is Christ, and to die is gain."                    |
|                                            --- Philippians 1:21 (KJV)   |
---------------------------------------------------------------------------
|   Ryan Kirkpatrick  |  Boulder, Colorado  |  http://www.rkirkpat.net/   |
---------------------------------------------------------------------------