Got a problem that I haven't been able to resolve with UK support - hoping
that someone on the list with knowledge of Unidata internals can help.

We have to use stopudt to kill sessions where a network connection has been
lost - can't use deleteuser because that requires NT administrator rights
(the users of our system don't have those, and IT people with those rights
don't have logins for our system).

Problem is that stopudt is successfully killing a process, NT is then
reusing that session's pid for a new session, and then stopudt is killing
that new session.

I'm guessing that stopudt does something along the lines of...

While [process id exists as unidata session] and [up to a maximuim number of
repeats] do
  Issue kill signal for process id

We've got some processes that run as phantoms that are [very] occasianally
getting killed off a few seconds after login.

I'm certain that this is caused by stopudt running a second hit (Tarantino
style) on a process number that it did kill but thinks it didn't.

It should be pretty easy for IBM to add a check to stopudt that the process
is still the same one (by checking the start time for example) but failing
that I need to put a sleep at the top of the phantom processes so that they
get killed while sleeping instead of while updating files.

In order to know how long to sleep for, I need to know how many loops
stopudt does and how long it pauses for at each iteration.  Can anyone help
with that?

Various UD versions from 5.2.9 to 6.0.7 - although problem only seen on
5.2.9 Win 2k site.


u2-users mailing list

Reply via email to