Need workaround for client-principal seal method as it crash

Posted by Darren Parr on 31-May-2013 12:42

Hi

In 11.2.1 on my PC, this crashes and creates a protrace when executing the seal() method.

Any ideas on how to get around it. The procedure below is an appserver connect procedure for a state-reset appserver running on my PC.

I have trust application domain registry set...

/* asconnect.p */

DEFINE INPUT  PARAMETER pcOperator AS CHARACTER NO-UNDO.
DEFINE INPUT  PARAMETER pcPassword AS CHARACTER NO-UNDO.
DEFINE INPUT  PARAMETER pcDomain   AS CHARACTER NO-UNDO.

DEFINE VARIABLE hCP        AS HANDLE    NO-UNDO.

 
CREATE CLIENT-PRINCIPAL hCP.


SECURITY-POLICY:REGISTER-DOMAIN("dom1","pass").
SECURITY-POLICY:LOCK-REGISTRATION().
 
hCP:INITIALIZE(pcOperator + "@dom1",?,?,pcPassword).
hCP:SEAL("pass").
 
MESSAGE hCP:LOGIN-STATE. /* <- It never gets here

Also here is some of protrace..

Exception code: C0000005 ACCESS_VIOLATION
Fault address:  0128F16C 01:0012E16C C:\DLC11\bin\_proapsv.exe

Registers:
EAX:00000041
EBX:00FB8E80
ECX:006CEA01
EDX:00000041
ESI:FFFFB2FC
EDI:00000000
CS:EIP:0023:0128F16C
SS:ESP:002B:006C6D60  EBP:006D0000
DS:002B  ES:002B  FS:0053  GS:002B
Flags:00210286


Debugging dll: C:\DLC11\bin\DBGHELP.DLL
Symbol Path:
C:\DLC11\bin;C:\DLC11\pdbfiles


Call Stack:
Address   Frame
0128F16C  006D0000  dbut_utSetNumFiles+BC85C

I could really use a workaround for this please and quickly....

Thanks

Darren

All Replies

Posted by andrew.thornton@redprairie.com on 31-May-2013 12:43

Sorry I missed your email. I am out of the office until Wednesday 6th June with little access to email.

Thanks, Andrew

Posted by Peter Judge on 31-May-2013 12:48

In your real code, do you call REGISTER-DOMAIN on every connect?

Was just trying in 11.3 and don't see a crash, but I do see the following messages when I run the code a second time. I wonder whether these errors are what's causing the crash.

Posted by Darren Parr on 31-May-2013 12:55

Hi

Thanks for your input.

Not on every connect. Heres a strange thing. Im using a 2 licence PDSOE and so I can only start 2 agents. I have a single connect statement from the client. IT actually connects twice to 2 different appservers. It obviously crashes twice..Somehow the broker must realise its happened on a connect and connects me again to another agent. The db has lots of wdog messages due to the crashes which is expected.

Its definitely a full blown crash. I did try adding the domain to the db and _extsso and usng it as a static domain and not using the session domain registry and that also crashes. I'm unsur ehow to get around this. I can't believe this hasn't been picked up.

Thanks

Darren

Posted by Darren Parr on 31-May-2013 14:11

Peter

Is your appserver state-reset? Mine is.

I think I expect to call this on every connect as it has session level scope and its a state-reset appserver. On disconnect the whole as connection is trashed and made available for the next connector.

-D

Posted by Peter Judge on 31-May-2013 14:18

No, state-free.

I would still say that you should only ever load the domains once per AVM/prowin/proapsv session.

-- peter

This thread is closed