[linux-tr] Mysterious Congestion Errors

[rokonis@us.ibm.com]

I have several Linux systems on the test lab token ring along with several dozen
systems of various flavors (OS/2, WINDOWS 95, AIX, etc).  An otherwise idle
Linux system will suddenly start experiencing congestion errors as determined by
the local lab administrator (shown on his WIN PHARAOH).  Performance on the ring
then goes to hell.  Disconnect the Linux box and all returns to normal.
Reconnect LINUX, then wait several weeks... The lab administrator can only tell
me there are congestion errors and packets dropped.  There are no messages on
the Linux box effected.  This has occurred at the 2.0.28 level and most recently
at 2.2.2.  We are using IBM 16/4 MCA  TR cards.  As a temporary "fix" I have
isolated all of my Linux boxes (except for one) on a separate ring.  This is not
a good solution for obvious reasons.  Any help appreciated.