AIX

Originally posted by: Jones Madruga

Hi,
I had an application that was running with pid 262384 and was listening under 7001 port.
My application crashed and i had to use kill -9 262384 to killed it.
When i use ps to see if the process died i can still see it.
server:/home/user1 $ ps -aedfklm | grep exiting
- A - 262384 - - - - - <exiting>
200001 A user1 286726 442490 0 60 20 16f917400 244 f10001003aa45c48 16:37:57 pts/1 0:00 grep exiting

And the port still listening.
server:/home/user1 $ netstat -Aan | grep 7001 | grep LISTEN
f1000200003a83b0 tcp 0 0 10.1.1.22.7001 . LISTEN
f10002000026b3b0 tcp6 0 0 ::1.7001 . LISTEN
f1000200001613b0 tcp 0 0 127.0.0.1.7001 . LISTEN

The only way that i found to really kill this process is rebooting the machine. I also tried rmsock, but if no effects.

I know that the problem is with my application but is it possible to recover AIX after that ?

Anyone can help me?

Thanks in advantage
Jones Madruga

Originally posted by: Jones Madruga

No ideas?
I'm still trying to find a solution.

thank you.

Jones Madruga

Originally posted by: SystemAdmin

A dumb question. If you leave the process for an hour, is the netstat entry still there? The reason I ask is that some listening processes have to stay around for a length of time (about 15 minutes by default) after the user has disconnected; just in case they reconnect. If you check the netstat in this period then you will find the entry is still there; after the timout period, the netstat entry is discarded.

Originally posted by: Jones Madruga

Yes, it still there even after some days...
So, what should I do?
thankz
Jones Madruga

Originally posted by: niella

Hi,

The exiting state indicates that the SEXIT flag is set for the process, and if it stays there it means that the operating system is unable to perform the necessary cleanup that would have enabled it to completely remove it from the process table. This could be due to various issues, for example a thread deadlock or race condition. It could likely be a problem relating to your version of AIX. If you run a command like instfix -ia|grep -i exiting" on a system with a recent version of AIX you might find various possible causes that relate to O/S defects.

Normally this does not present much of a problem, and you should only start worrying if there is a large number of them, or if they are causing tangible issues like making it impossible to unmount a certain file-system.

So my advice is to not do anything unless there are hundreds of these - apart from checking if there is a later version of AIX available for your system.

Regards,
Niel

Originally posted by: tschonnie

Hi Jones,

we had the very same problem on AIX 6.1 and were able to recreate and trace it for IBM. They found the problem and build APAR IZ72993.
The short answer from AIX Kernel Development:

It seems to be a pure timing
issue due to very high load. 2 threads were competing on the same
socket, so one thread released the socket and the other thread is
trying to access it.

So ask for APAR IZ72993 in a PMR.

best regards,
Johannes

Originally posted by: MrSatu

Hi Jones,

Did you manage to log an PMR and get a fix for this APAR?

Regards

Originally posted by: MrSatu

Hi Jones,

Did you manage to log an PMR and get a fix for this APAR?

Regards

Originally posted by: Pushkal

Guys,

can any one suggest which patch level/ fix pack level - IZ72993 was shippped.

Originally posted by: dukessd

http://www-01.ibm.com/support/docview.wss?uid=isg1IZ72993

Not available as an APAR yet, either subscribe for updates or call IBM and ask for an ifix.

Originally posted by: SystemAdmin

Hi,
I recommending a "kill -9"
A straight "kill" should be used first.
Wait a few minutes and give the system time to gracefully kill the process.
If that does not work a "kill -9" can be done, but be aware that it may
leave zombie processes because it kills without closing applications down
appropriately.
In some cases, you can also "Hup" a process. This is also a graceful
shutdown of a process, but automatically restarts the process. If it is
successful, the date/time in the "ps -ef" will be current.
regards,
G prasad babu
Networkk solutions,chennai

Originally posted by: john1212

could someting still give to the port 7001
can stop its?
You can disconnect the wire ethernet.

Originally posted by: aschauhan

Hi Everybody,

I am using AIX 5.3 TL9 and my java process  has gone in exiting state and socket 38080 will be not be released until exiting processes to be cleaned up. Forcefully kill -9 won't work on it. Can someone please help me on this, I don't want to reboot my system.

1. Status of java process in kernel database.

SLOT NAME     STATE      PID    PPID          ADSPACE  CL #THS
pvproc+017800   94*java     ACTIVE 385192 0000001 000000045F31F590   0 014D

2. Output of ps -ef

root@[brippu9001]/# ps -ef|grep -i 385192
       - 385192      -   -                     - <exiting>