I have a linux guest under VM 4.2 (SLES 7 - Patch CD1 - k_timer 2.4.7-9)

The last few days, we have seen on three occasions were the box is
partially hung.  I can ssh in fine, I can cat/vi files no problem,  but
any process related commands (ps, top, etc) hang without any response
after pressing enter.   Procinfo shows the box is 96-98% idle at the time
but more importantly we see the context field in procinfo incrementing
wildly (Peak rate of about 50,000,000 switches per minute!).   Netstat
will work, however if you include the -p switch to see what processes are
associated with the network connections it will also hang.  It doesn't
ever seem to recover from this situation, and the only way we have found
to get out of it is to logoff the guest from VM, which causes as big a
mess as one would expect from having to hit the virtual recovery switch.

The only application on this box is a java based application which uses
RMI connections between the clients and DB2/UDB.

Any suggestions on how to help narrow down what process(s) specifically
are going haywire would be appreciated.  I had seen somewhere in the past
there were ways to "peek" into a running linux guest from VM to see what
was going on in there, if anyone can point me to some type of recipie on
doing that I would appreciate it.

Any other suggestions on where to look or what to do is also appreciated

Thanks in advance
Jeremy Warren

Reply via email to