Intermittent Login Problem

[gbarcalow]

When logging into OpenVMS users will sometimes experience an error where; after entering their username and password will be immediatly disconnected from the server.  The last info transmitted from the server before disconnect is this:

HP OpenVMS Industry Standard 64 Operating System, Version V8.4

Secones/Minutes/Hours later they will try again and connect just fine. The error never seems to occur for the SYSTEM account so thats a plus. When it works our users get the full login text:

HP OpenVMS Industry Standard 64 Operating System, Version V8.4
Last interactive login on Monday, 1-OCT-2018 14:11:40.18
Last non-interactive login on Monday, 1-OCT-2018 14:10:19.50
**** Logon authenticated by LDAP ****
OpenVMS password has been synchronized with external password

Our setup

We are using ACME Ldap for authentication. The problem occurs both with LDAP authenticated users and native OpenVMS users stored in the SYSUAF.DAT. I know the LDAP connection is fine, because if someone types the wrong password, they get an invalid login message, and when it's typed correctly it will either let them in or disonnect them depending on the whim of the machine.

Everything was working perfectly last week, but we moved offices this past weekend and now this issue has cropped up. I have rebooted the entire server to no great success.

Any ideas from people with more OpenVMS experience than I? I don't even know where to look for exceptoions, login errors, or anything. 

Your time is greatly apprecitated in helping me resolve this issue.

[TonyMcGrath]

You might something helpful in the SYS$MANAGER:ACME$SERVER.LOG file, though the contents can be a bit cryptic. If there's an issue with the connection to your LDAP server you'll find the details in there.

[gbarcalow]

So the problem has to do with interactive logins I've found.

set logins

show that I have 129 or 128 available logins. I only have 40 users, so I thought it odd. So, I bumped the number and users were able to connect. But, that didn't solve the problem, the number just continuted to climb out of control all the way to the new number within about an hour. All the users are coming from a server that users CONNX to query our RMS database. When I do a show users /full all the users from that server are disconnected. I don't know why the system, is all of the sudden, not sweeping or expiring those.

[abrsvc]

If you have accounting turned on, look at the logout records and see what the final status is.  Perhaps the logout is not complete and some structures are hanging around that shouldn't be there.

You may need to take a look using ANALY/SYS too.

Dan

[Brad McCusker]

You've likely got a bunch of those application processes being left behind on the system.  you've got to do some analysis on them to figure out what they are waiting for, why aren't they going away.   For some reason they aren't getting disconnected.   Analyze/system is likely to be your friend here.

Brad McCusker

Software Concepts International

www.sciinc.com