We started with vSwitch grants in SYSTEM CONFIG, then moved to reading lists
of Linux guests and dynamically granting them to the vSwitch. We then
switched (no pun inteneded) to COMMAND statements in each CP Directory to
grant the vSwitch and couple the NIC to the vSwitch. This seemed to be the
most consistent and least intrusive method of emulating an open vSwitch, and
reduced the need for adding users or config lines to other files.

Except for one slight drawback: If the two COMMAND lines are in the user¹s
CP Directory entry, then the grant is done, followed by the couple, and
everything is happy. If the same two COMMAND lines are in a profile,
included in the user¹s CP Directory entry, then it appears that either the
commands are executed in the wrong order, the grant never gets executed, the
grant executes but with the wrong or no userid, or the system doesn¹t allow
the grant to complete before attempting the couple. In any case, the user
does not get a working NIC coupled to the vSwitch.

Could someone verify this behavior on their system? I¹m getting ready to
open an ETR with IBM, but I¹d like verification that I¹m not overlooking
something else...

The minimal entries needed to recreate this are:

TSTDFLT DIRECT:

PROFILE TSTDFLT
COMMAND SET VSWITCH VSWG GRANT &USERID
COMMAND COUPLE 8200 TO SYSTEM VSWG
*
MACH XA
CONSOLE 0009 3215 T
NICDEF 8200 TYPE QDIO LAN SYSTEM VSWG

-----
TESTUSER DIRECT:

USER TESTUSER XXXXXXXX 1G 2G G
INCLUDE TSTDFLT
*
*
*  COMMAND SET VSWITCH VSWG GRANT &USERID
*  COMMAND COUPLE 8200 TO SYSTEM VSWG
*
*  CONSOLE 0009 3215 T
* NICDEF 8200 TYPE QDIO LAN SYSTEM VSWG

Try them this way; The user seems to never get granted. Then comment out the
INCLUDE and uncomment the other four lines. The user gets granted the first
attempt.

Any comments or ideas?

-- 
Robert P. Nix          Mayo Foundation        .~.
RO-OE-5-55             200 First Street SW    /V\
507-284-0844           Rochester, MN 55905   /( )\
-----                                        ^^-^^
"In theory, theory and practice are the same, but
 in practice, theory and practice are different."




Reply via email to