[9012] in Athena Bugs
rt 7.3P: RT login procedure in cluster m2-225
daemon@ATHENA.MIT.EDU (gptesler@Athena.MIT.EDU)
Mon Feb 24 13:05:36 1992
From: gptesler@Athena.MIT.EDU
To: bugs@Athena.MIT.EDU
Date: Mon, 24 Feb 92 13:03:45 EST
System name: m2-225-2
Type and version: RTPC-ROMPC 7.3P
Display type: apa16
What were you trying to do?
Login
What's wrong:
It took 16 minutes to login (and in the past has often taken 10+
minutes on the machines in m2-225). Other people in the room were
reporting similar problems, and warning anyone new who came in the
room about the situation.
Here is what was shown in the Console window when I tried to login.
12:19 Athena Workstation (RTPC) Version 7.3P Tue Oct 22 16:45:55 EDT 1991
12:20 Setting up environment...
12:21 attach: /afs/sipb.mit.edu/project/sipb linked to /mit/sipb for filesystem sipb
12:21 douzzer: Timeout while contacting mount daemon on CYRUS.MIT.EDU
12:21 attach: Ignoring host CYRUS.MIT.EDU for filesystem douzzer due to previous host errors
12:22
12:22 User gptesler over disk quota on /mit/gptesler, remove 165K within 6.2 days
12:22
12:22 Starting mwm window manager...
12:23 Starting the Athena Dashboard...
12:23 Creating logout button...
12:23 Creating initial xterm window...
12:23 Running standard startup activities...
12:26 Running custom startup activities listed in ~/.startup.X...
12:29 znol: No such file or directory while getting WindowGram port
12:29 Session initialization completed.
At this point, there was still no dash menu, Logout button, mwm
decorations, or xterm window. At some point, the dash menu and
Logout button appeared. At some later point, the Console window
and Logout button were reparented and decorated by mwm. At 12:35
(by the Dash clock), the xterm window finally appeared.
Prior to the "Session initialization completed" message, the
cursor was an "X" outside the Console window and an arrow inside
it; after this message, it was an hour glass. The transition may
not have been exactly when the message was displayed, but it was
approximately then.
What should have happened:
I presume it's intended that the login procedure be somewhat faster.
I know the RTs are slower than some of the other computers, but
even in other clusters they're not THAT slow... Are there poor
network connections to m2-225, or some timing peculiarity in the login
procedure that, due to conditions in m2-225, causes the machine
to be trying to do too many things at once with the result that it
gets none of them done?
Perhaps there should be a method to allow cleanly aborting the login
procedure, such as simultaneously pressing some strange combination
of keys. Of course, such a feature must somehow be documented,
perhaps by means of a poster in cluster m2-225 and any other
clusters which exhibit similar login time performance problems.
Putting the Dash Logout button up early is not a sufficient
solution, because (1) it did not appear until sometime after 10
minutes into the login procedure, and (2) the cursor was an hourglass
for the whole time it was up until the xterm window appeared, so
when I pressed it around 12:34, it was ignored.
Please describe any relevant documentation references:
