Hangup signal not send after AIX version upgrade (6100-09-03)

Hi ! I am a new "accidental" DBA since the last one is no longer working for us. Here is my issue, last weekend our It departement has updated our AIX server to version 6100-09-03 and now it seems like Progress doesn't received a Hangup signal when user close the telnet session using the X. We have both CHUI and GUI enviroment, but we only have this issue from the CHUI one. The Progress version installed is 10.2B07. Any clue on where we should look at ? We think that this might be a Java issue because the AIX upgrade as update it, but at this time we are not sure and we are doing some test. Maybe a setting on AIX has changed and we need to change a Progress setup...

Thanks !
Bye

When you say upgraded.... was this upgraded in place? Meaning this is the exact same hardware as before or do you have access to the old AIX version and hardware...

This isn't very likely to be a Progress issue and is more likely an AIX setup/termnal emulation issue.

JAVA isn't going to cause this issue because it is only used to manage the name server, application servers, etc.

Yes, this is the exact same hardware as before. I am pretty sure it's a AIX setup that have changed, but I Im in the blind. I will probably open a case at Progress just in case

Or maybe someone here is an expert in AIX

Maybe a trap change in /etc/profile or in the users .profile?

What exactly is happening now and what do you think was happening before?


Progress KB about traps...

http://knowledgebase.progress.com/articles/Article/000027890

Not a fan myself and I prefer to clean up the sessions myself.

When the Telnet session is closed by the X, the user don't get disconneted from the database. In the database log, I see the line "vv_flush:I/O error 5 on fd 1", but I don't see the "Hangup signal received" as prior. So we end up with a lots of dead connections and using "nmon" in unix I see a lots of _progress process , taking a lots of cpu and I have to disconnect them manually. Thanks for your information, I will take a look at the Progress KB and take with the Unix guys about the trap

we had a similar issue with Solaris many years back. If I recall correctly, we changed the default setting of the TCP Keepalive parameter. make sure you are running a watch dog process.
I wrote a script that tailed the db log looking for the word vv_flush and if found it disconnected the user. That kept the server load down.

The problem was on AIX shell, after installing the following service pack 6100-09-04-1441, everything works as expected ! Thanks for your supports guys !