I suspect the LOGON clist is different between the 2 images. Check the parm
field of the logon proc.
In particular, the logon clist on LPAR1 is different than LPAR2 (content, not
name).
The LPAR1 clist probably creates a dataset if not present. The LPAR2 clist
does not.
This may also be due to a "lpar specific" clist higher in the concat.
e.g. SYS1.LPAR.CLIST
SYS1.SYSPLEX.CLIST
HTH,
<snip>
We have 2 systems in a basic sysplex, sharing everything: RACFDB, JES2, Master
catalog, Parmlibs etc. The only differences between the sytems is that they
are each on a different CEC, and one system has a few licensed products that
the other doesn't such as MQ, Syncsort, Qwikref.
We define a new user to RACF, and all options defined on TSO segment. New
user can logon to system 1 successfully. Then logon to system 2 successfully.
Everybody is happy. If the new user tries to log onto system 2 *first* the
logon fails with:
....snippage
IF the user goes back and logs onto system #1, it is successful. THen back to
system #2, now it's successful too.
There are very few differences between these systems. Everyone uses RACF for
TSO logon, the only thing in uads is IBMUSER0. SYS1.UADS is catalogged in the
shared master catalog on a different volume, Z13RES is the currently IPLed
sysres volume. MSTJCL00 contains:
//SYSUADS DD DSN=SYS1.UADS,DISP=SHR
I can't find anywhere uads is referenced by volser, or ****** or &SYSR1 or
anyting. I cannot explain why initial logon works on one system and not the
other. Does anybody have ideas of what else to check? I've about run out of
things to look at.
</snip>
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
