EDC5157I An internal error has occurred

General discussion of the Co:Z Toolkit
Post Reply
rmitchell
Posts: 6
Joined: Thu May 08, 2008 10:43 am
Location: Raleigh, NC

EDC5157I An internal error has occurred

Post by rmitchell »

Installed the toolkit on z/OS V1R8 and Windows XP. SSH was previously installed on the z/OS system. Can SSH from z/OS to windows workstation and vice versa ok...When running the batch launcher receiving the following message.

spawnp(/bin/ssh) - EDC5157I An internal error has occurred

No messages on the console indicating any RACF problems. Have run the batch job with trace. Can send if needed? Doesn't really tell me much more about the problem. If I could get the return or reason code then progress might be made.
dovetail
Site Admin
Posts: 2022
Joined: Thu Jul 29, 2004 12:12 pm

Post by dovetail »

It could be something simple like your effective region size is too small. The default region size in the COZPROC is 32M -- are you limiting in your job card or via a an installation exit?

What I can suggest is that you set the z/OS environment variable that adds errno/errno2 information to the EDC message that you are getting.

One way to do that is to add this to your job:

//COZCFG DD *
server-env-_EDC_ADD_ERRNO2=1

This should probably be a default setting in the launcher.
Give it a try and post back the error message which should contain the errno and errno2 that we can probably figure out what is preventing ssh from spawning.
rmitchell
Posts: 6
Joined: Thu May 08, 2008 10:43 am
Location: Raleigh, NC

EDC5157I An internal error has occurred.

Post by rmitchell »

Region size is set to 0M. After adding the environment variable recommended ,received the following message and reason code:

spawnp(/bin/ssh) - EDC5157I An internal error has occurred. (errno2=0x0B1B0473).

The 0473 indicates that the process could not be spawned. Not a lot of info for the reason code available, either from IBM service link or the internet. Must be something that I have not setup correctly, but finding out what it is a different story? I also added a heap storage environment variable for BPXROOT, but no help.
dovetail
Site Admin
Posts: 2022
Joined: Thu Jul 29, 2004 12:12 pm

Post by dovetail »

I'm not sure what the problem is, but I can suggest a couple of things that might help pin it down:

1) Verify that you can login to a tty Unix shell (ssh or telent) and try the ssh command using the same userid that you are using in the batch job.

2) With the same userid, run the DTLSPAWN sample proc, launching ssh

// EXEC DTLSPAWN
//STDIN DD *
env
ssh user@host env
//

3) Use DTLSPAWN to launch ssh directly without a shell:
// EXEC DTLSPAWN,CMD='/bin/ssh user@host env'

If it works from with a shell but not directly, maybe you are setting some environment variables in your profile that are affecting it.... post the local env output and that might give us a clue.
dovetail
Site Admin
Posts: 2022
Joined: Thu Jul 29, 2004 12:12 pm

Post by dovetail »

I still think that there is a good chance that your region size is the problem. Perhaps it is being limited by an installation exit?

If you run DTLSPAWN, with the same job card as you are running COZPROC, but add the proc keyword parm LOGLVL='-LD' which will print debugging messages showing the effective region size above and below the line. (We will probably add this feature to the next release of COZ).
rmitchell
Posts: 6
Joined: Thu May 08, 2008 10:43 am
Location: Raleigh, NC

EDC5157I An internal error has occurred

Post by rmitchell »

I am able to ssh from the z/os system to the windows and vice-versa. I ran the DTLSPAWN job using both suggestions. Also ran it with LOGLVL='-LD' . The results were the same 01b10473. The LOGLVL parm did not produce any data indicating region sizes?
rmitchell
Posts: 6
Joined: Thu May 08, 2008 10:43 am
Location: Raleigh, NC

e

Post by rmitchell »

XXDTLSPAWN PROC LOGLVL='', < -LD = debug, -LT = trace
XX CMD='', < command and args to spawn
XX LIBRARY='SYS1.COZ.LOADLIB', < STEPLIB FOR DTLSPAWN module
XX REGSIZE='32M', < EXECUTION REGION SIZE
XX LEPARM='' < Option LE runtime parms
4 XXSPAWN EXEC PGM=DTLSPAWN,REGION=&REGSIZE,
XX PARM='&LEPARM/&LOGLVL &CMD'
IEFC653I SUBSTITUTION JCL - PGM=DTLSPAWN,REGION=64M,PARM='/-LD '
5 XXSTEPLIB DD DSN=&LIBRARY,DISP=SHR
IEFC653I SUBSTITUTION JCL - DSN=SYS1.COZ.LOADLIB,DISP=SHR
6 XXSYSOUT DD SYSOUT=* < DTLSPAWN and system messages
7 XXSTDENV DD DUMMY < can be used to set env vars
8 //STDIN DD *
X/STDIN DD DUMMY < stdin to spawned pgm (may override)
9 XXSTDOUT DD SYSOUT=* < spawned program STDOUT
10 XXSTDERR DD SYSOUT=* < spawned program STDERR
11 XX PEND
STMT NO. MESSAGE
2 IEFC001I PROCEDURE DTLSPAWN WAS EXPANDED USING SYSTEM LIBRARY CPAC.PROCLIB
ICH70001I SERROM LAST ACCESS AT 13:30:41 ON MONDAY, MAY 12, 2008
IEF236I ALLOC. FOR DTLSPAWN SPAWN
IEF237I 1175 ALLOCATED TO STEPLIB
IEF237I JES2 ALLOCATED TO SYSOUT
IEF237I DMY ALLOCATED TO STDENV
IEF237I JES2 ALLOCATED TO STDIN
IEF237I JES2 ALLOCATED TO STDOUT
IEF237I JES2 ALLOCATED TO STDERR
IEF142I DTLSPAWN SPAWN - STEP WAS EXECUTED - COND CODE 0126
IEF285I SYS1.COZ.LOADLIB KEPT
IEF285I VOL SER NOS= Z8SYS1.
IEF285I SERROM.DTLSPAWN.JOB00029.D0000102.? SYSOUT
IEF285I SERROM.DTLSPAWN.JOB00029.D0000101.? SYSIN
IEF285I SERROM.DTLSPAWN.JOB00029.D0000103.? SYSOUT
IEF285I SERROM.DTLSPAWN.JOB00029.D0000104.? SYSOUT
IEF373I STEP/SPAWN /START 2008133.1335
IEF374I STEP/SPAWN /STOP 2008133.1335 CPU 0MIN 00.22SEC SRB 0MIN 00.00SEC VIRT 88K SYS 288K EXT 6856K SYS 12944K
IEF375I JOB/DTLSPAWN/START 2008133.1335
IEF376I JOB/DTLSPAWN/STOP 2008133.1335 CPU 0MIN 00.22SEC SRB 0MIN 00.00SEC
rmitchell
Posts: 6
Joined: Thu May 08, 2008 10:43 am
Location: Raleigh, NC

Post by rmitchell »

Output from env follows:

_EDC_ADD_ERRNO2=1
MAIL=/usr/mail/
PATH=/usr/local/bin:/usr/sbin:/usr/lpp/zWebSphere/V6R1/java/J5.0/bin:/usr/lpp/Printsrv/bin:/bin:.
PS1=$LOGNAME:$PWD: >
_BPX_SPAWN_SCRIPT=YES
_=/bin/env
LANG=C
LIBPATH=/usr/lpp/Printsrv/lib:/lib:/usr/lib:.
_BPXK_JOBLOG=STDERR
_BPX_SHAREAS=MUST
HOME=/
TZ=EST5EDT
MANPATH=/usr/lpp/coz/doc/man/%L:/usr/lpp/Printsrv/man/%L:/usr/man/%L
NLSPATH=/usr/lib/nls/msg/%L/%N.cat:/usr/lpp/Printsrv/%L/%N:/usr/lib/nls/msg/%L/%N
coz
Posts: 391
Joined: Fri Jul 30, 2004 5:29 pm

Post by coz »

I'm not sure why DTLSPAWN isn't working for you - you should always be
getting a copyright message displayed on DD:SYSOUT. The RC=126
also isn't anything that I recognize. Do you see *any* debugging
information on DD:SYSOUT when you run it?

I was mistaken however about the region size information - that
feature isn't in the released version which explains why you don't see
it. We have just released a new version (1.0.7) of Co:Z for z/OS
this afternoon that adds REGION size debugging messages to both the
DTLSPAWN and COZLNCH programs.

I would like to double check something with your /bin/ssh executable.
Please verify that you don't have the "s" bit set for either the user
or group and that it is not APF authorized. You should see something
like this:

$ ls -al /bin/ssh
-rwxr-xr-t 2 XXXXXX YYYYYY 1531904 Mar 8 2007 /bin/ssh

$ extattr /bin/ssh
/bin/ssh
APF authorized = NO
Program controlled = NO
Shared address space = YES
Shared library = NO


The new version also has a new property "ssh-shareas" which defaults
to "MUST" but can also be set to "YES". Since it is not strictly
necessary to run the ssh client for Co:Z in the same address space,
this might be a workaround for you until we figure out why ssh will
not spawn in the same address space.
rmitchell
Posts: 6
Joined: Thu May 08, 2008 10:43 am
Location: Raleigh, NC

Post by rmitchell »

The bits on the ssh executable were set as apf authorized and program controlled. Turned the bits off and was able to use DTLSPAWN. The cozserver should also now work.

Will check to see why the bits on the ssh module were set to apf authorized and program controlled?

Thanks for your help,,,Roy
dovetail
Site Admin
Posts: 2022
Joined: Thu Jul 29, 2004 12:12 pm

Post by dovetail »

There is no reason that I can think of that /bin/ssh should be either APF authorized or program controlled. It doesnt' ship from IBM that way. I'm glad we could resolve the problem. We will try to add this to our FAQ..
Post Reply