[411] in Athena Bugs

home help back first fref pref prev next nref lref last post

OLC and bldg 9

daemon@ATHENA.MIT.EDU (Thomas J. Coppeto)
Fri May 27 17:14:06 1988

To: hotline@ATHENA.MIT.EDU, consultants@ATHENA.MIT.EDU,
Cc: beth@ATHENA.MIT.EDU, olc-bugs@ATHENA.MIT.EDU, geer@ATHENA.MIT.EDU,
Date: Fri, 27 May 88 17:12:53 EDT
From: Thomas J. Coppeto <tjcoppet@ATHENA.MIT.EDU>

Hello.

There is something wrong with communication from the e40 and other 
public Athena clusters to some of the bldg. 9 cluster of IBM RT's,
namely penn.mit.edu and bullfinch.mit.edu. 'Finger' and 'write' seem
to hang when trying to contact these machines. Other bldg. 9 machines
seem to be ok but no one was on them. Problems seem to arise when someone
is logged in. I spoke to Joe Ferreira (ferreira, x3-7410) and he said 
that this must have been something new since it has worked in the past.

The major problem for us is that OLC depends on a successful write
to the user. Due to some problems in the code, when the write fails
due to a time out, the consultant is not fully inserted into the
'consultant ring'. This is a just a theory but the real result is that
the OLC daemon seemingly hangs and soon crashes and all current OLC 
conversations are lost.

The OLC server is admidst a revision but it be a while (until the next
Athena release) until the problem is solved in the field. I would    
really like to know what causes these problems in communication, if it
is known. These problems may have accounted for a fair perecntage
of OLC problems in the past if other machines have experienced similar
problems. 

In the meantime I request that users of the building 9 workstations
not use OLC until the problem is solved. If a user from that area 
asks an OLC question, (OLC consultants:) do not grab it without trying 
to write or finger the user outside of OLC. The danger is that we may
disrupt service for other Athena users in need.

I hope this problem is a transient thing and I sincerely apologize
to users of these workstations. I hope to hear from Athena operations
as to the cause and possible short term solution to the problem.

						Tom Coppeto
                                                OLC maintainer

home help back first fref pref prev next nref lref last post