I somewhat agree with you Alan (and these minidisks are indeed already
in the HCPRww so that CP doesn't even consult RACF for a RR link.
As for these passwords: there is no real convention thet the passwords
msut be "monyyyy", they can be anything that explains the
content/level. It would be more secure without them, but I prefer
knowing what level is on an MDISK and avoid VM (or VM application)
outages due to mistakes as I consider the risk of someone deactivating
RACF (or starting a CP without RACF) much smaller.

You can not my vote to extend the MDISK statement with a COMMENT field
that DIRMAINT will always keep with the MDISK card.  e.g. a C-like
notation (and waw more than 3 words as comment, and no prereq for an
ESM)
  MDISK 019E 3390 1735 190 VTE521 MR ALL // Fixes for CA products DEC2007

2008/2/8, Alan Altmark <[EMAIL PROTECTED]>:
> On Friday, 02/08/2008 at 10:06 EST, Kris Buelens <[EMAIL PROTECTED]>
> wrote:
> > If you are lucky enough to have an ESM like RACF, what means the LINK
> > checks are not password based, you can work the way I set things up
> > for my client: use the minidisk passwords as descriptive comments.
> > Here an extract of MAINT's entry:
> > MDISK 0490 3390 2362 107 VTERES RR ALL ZVM520 NOV2006
> > MDISK 0493 3390 1433 167 VTERES RR ALL ZVM520 NOV2006
> > MDISK 019D 3390 681 146 VTEBKP RR ALL ZVM520 NOV2006
> > MDISK 049E 3390 2687 190 VTERES RR ALL ZVM520 NOV2006
> > MDISK 0CF1 3390 587 80 VTERES RR
> > MDISK 0CF2 3390 408 80 VTEBKP
> > MDISK 0190 3390 399 107 VTE52R MR ALL ZVM520 AUG2007
> > MDISK 0193 3390 0001 167 VTE521 MR ALL ZVM520 AUG2007
> > MDISK 019E 3390 1735 190 VTE521 MR ALL FIXESCA DEC2007
> > Side remark:
> > As one can see, we set the read password to ALL for these "general
> > public" minidisks.  This way if we'd be forced to IPL without RACF,
> > people would still be able to LINK RR without password (but, even
> > though I still have a CP nuc without RACF on CF1, we never had to use
> > a CP-without-RACF in the 19 years we run with RACF
>
> Since they are public and you don't require auditing of their access, you
> can add them to RACF's GLBLDSK macro to prevent RACF from protecting or
> auditing read-only access.
>
> Unfortunately your use of the password fields in that way creates a
> security exposure.  If the system does come up without RACF, OR if the
> VMMDISK were deactivated, anyone who knows your naming convention can get
> r/w access.  If, in 19 years, you never had to run without RACF, then I
> think the risk of the security exposure exceeds the risk of a RACF outage
> sufficient to make you run without RACF.
>
> Alan Altmark
> z/VM Development
> IBM Endicott
>



-- 
Kris Buelens,
IBM Belgium, VM customer support

Reply via email to