UCX : rsh freeze between a mainframe and a VAX

[Stéphane Boneff]

Hello,


I have a problem with RSH of UCX 4.1 eco 10 on VAX VMS 6.1.
(we cannot upgrade VMS).

I use it for communication between a VAX and a IBM3270.


In some situation, the IBM3270 can not execute
the RSH (it wait for 1 min and send the return code
"Foreign host aborted the connection 8556").

after n retry, it can execute the RSH.

But, when the pb occurs, the same host can execute a RSH to the same VAX in a another
JOB.

(in normal situation, there is a lot of succefull rsh from IBM to the VAX every day)

To stop and restart UCX
weekly (the VAX is shutdown only once a year)
don't work.

--> Except upgrading, is there another solution ?

Thanks.

[Wim Van den Wyngaert · Accepted solution]

Did you check the maximum number of allowed rsh ?

ucx sho servi rsh /fu

and compare limit with peak

Wim

[Peter Zeiszler]

Do you see any errors or failed connections on the VMS system when the IBM is attempting is RSH?

Does the RSH create and lock specific files during its initiation phase of the login? We have had RSH perform "logging" and if a second RSH of the same program was initiated it would not work even though it made it to the VMS.

I don't know if you need to check the account that the RSH is also attempting to connect to on the VMS. The VMS account might have a limited number of processes that it can create/use at the same time.
MC AUTHORIZE SHOW (Account name).
Look for the PRCLM. I don't know what error would be generated if that value is too low.

[Stéphane Boneff]

In VMS, when the pb occurs (that can be during 1 hour), there is nothing on VMS.
No intrusion, no error in accounting.

The RSH from IBM3270 is just a Submit command so it takes only few secondes to perform.


Prclm: 20
for then account name.


For ucx :
Service: RSH
State: Enabled
Port: 514 Protocol: TCP Address: 0.0.0.0
Inactivity: 5 User_name: Process: UCX$RSHD
Limit: 3 Active: 0 Peak: 2

peak is 2 and limit is 3. Since the last UCX restart, the pb did not occur.

I ask to my dec support to change the valeur of the limit and we will see.

(Our pb started in begining of 2005. For some systems, we were able to upgrade to UCX 4.2. Our dec support think about a bug in our 4.1 version.)

Thanks.

[Stéphane Boneff]

Limit of rsh was the problem. On the test system, i could simulate the pb.
and increasing this limit (support DEC said to increase both RSH and REXEC) has changed
the beavior.

Thanks!

one year !!!!

[Stéphane Boneff]

thanks to help me resolving this 1 year long problem !!

[Wim Van den Wyngaert]

Just a sidenote : you should know how many application you can block simply by doing some "rsh wait 01:00" to it (I once did that to see if the rsh executer was handling errors).

Wim

[Stéphane Boneff]

I exactly did that !
3 jobs "wait nnn"
1 job "DIR"

the job DIR has been rejected.