Sudden jump in cpu after a client change

[David Collier-Brown]

On Thu, Jan 29, 2004 at 02:54:42PM -0500, David Collier-Brown wrote:
 >   On a Samba 2.2 site, a colleague of mine observed a sudden
 > jump in CPU usage after a change made to the clients.
 >
 >  "A tech modified the XP workstations to forcible reconnect any
 > disconnected drive through a script that utilizes the WMI event.
 > As of Tuesday we noticed a considerable jump in CPU utilization
 > of our host."  When my colleague stops the smbd and the nmbd process
 > the server returns to about 27% utilization.
 >
 >   Any suggestions on identifying what just DOS'd us, and
 > what to do to mitigate it?

   Well, this turned out to be a client program which looped
to check that drives were mapped on client machines, and
checked every five seconds by
- looking for changes in mappings,
- enumerating all network drives
- reading the top-level directory
- mapping the drive it the read failed.

   Apparently there was some form of unreliability
with keeping drives mapped: I'll inquire further.

   Question for the samba team: what's the best way for a
client to keep it's drives connected during periods
of instability with their servers or network?  My
gut feel says "set the disconnect timout larger" (;-))

--dave
-- 
David Collier-Brown,       | Always do right. This will gratify
Sun Microsystems,          | some people and astonish the rest.
Toronto, Ontario,          |                      -- Mark Twain
(905) 415-2849 or x52849   | davecb at canada.sun.com