T.R | Title | User | Personal Name | Date | Lines |
---|
3427.1 | idle process killer? | VINO::MCARLETON | Reality; what a concept! | Wed Oct 03 1990 17:32 | 5 |
|
Does the customer run an idle process killer that might be killing
the WSA_ process?
MJC
|
3427.2 | | CSC32::FORSMAN | Ginny Forsman 522-4731 CSC/CS | Wed Oct 03 1990 18:55 | 6 |
| >>Does the customer run an idle process killer that might be killing
>>the WSA_ process?
Nope.
|
3427.3 | CPU time limit? | PEACHS::BELDIN | | Fri Oct 05 1990 14:49 | 12 |
|
How about a cpu time limit on the account? The way that DECwindows
works until 5.4 is that the system account is running in wsan and
then when you log in, the wsa process changes its uic to yours, but
you get charged the cpu time for the time that system spent blinking
the cursor. If that has exceeded your cpu time limit, out you go -
no session manager. I believe that the SYSTEM account doesn't
have this problem...
Rick Beldin
VMS Workstation Support
Atlanta GA
|
3427.4 | No CPU Limit on the account | CSC32::FORSMAN | Ginny Forsman 522-4731 CSC/CS | Wed Oct 10 1990 13:31 | 8 |
| Thanks for the suggestion, .3, but the CPU Limit on the users account
would only come into play after the user logs in. He would log in,
and the session manager would fail to start if he had exceeded his
cpu time limit.
In my customers case, the login box is disappearing, the WSA process
is terminating, we never get a chance to log in, as we don't
have a login box.
|
3427.5 | | CLTMAX::dick | Schoeller - Failed Xperiment | Wed Oct 10 1990 14:47 | 11 |
| I believe that .3 is slightly wrong.
I vaguely remember a note a while back that said that when you logout, the
process running the login box is using resources against your account (not
a system account) when someone new logs in, his account takes over. If
there is some account quota (CPU is a good one and there may be others) on
the account of the person who last logged out, then the login box might fail
over the weekend. The login box does use CPU because it is flashing the
cursor.
Dick
|
3427.6 | probably related to PQL_* parameters... | DECWIN::LEMBREE | Just do it. | Thu Oct 11 1990 12:21 | 10 |
| I think Dick's idea is on the right track, but I'm almost certain
that the new login process is a completely new process, not a remnant of
the old session. My guess would be that the customer's changed some PQL
parameters in SYSGEN, and that they're being used in the creation of the
new login process. When the login process runs up against some limit
imposed by PQL parameters, it'll probably be fatal. Of course, it'll
probably be caught and you'll get the final process status you're getting.
Like (I think) Dick said, while the machine's waiting for someone to log
in, it's not idle, it's busy doing little things like blinking the cursor,
so it's using real resources.
|