T.R | Title | User | Personal Name | Date | Lines |
---|
1760.1 | This was discussed and solved (?) back in April. | CGHUB::JREGAN | | Wed Nov 06 1991 11:31 | 91 |
| I acquired this from Mike. Evidently this problem was to be fixed,,
MCC EASYNET TECHNICAL PROBLEMS AND SOLUTIONS 4/30/90
MICHAEL KILLIAN
CTE ENGINEERING
DNRS 4+ will configure access control on Directories/Objects/Softlinks
relating to the existance of Node names in the EASYnet namespace. I have been
meeting with different groups (Session Control, MCC, etc) to determine the
specific access they need configured. I met and discussed MCC access
requirements with Kathy Nelson in LKG.
While talking with Kathy, we uncovered certain problems which, if left
unchanged would cause MCC to fail on EASYnet. This memo is intended to document
these problems and in the near future, their solution. MCC considers EASYnet to
be a valuable customer, and is anxious to have MCC meet the needs of EASYnet.
PROBLEM DESCRIPTION
1 had to do with naming objects in DEC:.DNA_NODE dir and
it has been fixed.
================================================================================
2 For Phase 4 names MCC is not tolerant of the existance of
backtranslation softlinks! If backtranslation softlinks exist
MCC will fail.
MCC is not tolerant of the existance of synonym softlink! If
synonym softlink exists then MCC will fail.
On EASYnet prior to MCC running DNRS4+ intends to load the
complete nodename namespace ( Synonym softlinks, Nodename
objects and Backtranslation softlinks ).
*indeed this has been done.
MCC with the command REGISTER NODE4
* attempts to load the nodename object If it already
exists then continue.
* attempts to load the backtranslation softlink
If it already exists then MCC fails.
The assumption here is that the node was previously
registered under a different name, and MCC has just detected
an illegal second registration (2 names 1 address !).
* attempts to load the synonym softlink
If it already exists then MCC fails.
(NOTE: DNRS4+ is also intolerant of the existance of
Backtranslation softlink and Synonym softlink.)
(NOTE: The MIGRATION TOOLS (Stan Goldfarb) IS TOLERANT of
the existance of Backtranslation softlink and Synonym
softlink. )
SOLUTION DNRS4+ and MCC have agreed to the following solution.
Consider the "REGISTRATION" of node information to work with
a set of data (Synonym softlink; Node object and Backtranslation
softlink).
When an attempt to add a Backtranslation softlink finds that
the softlink already exists. The application traverses the
softlink and finds what it points to.
The softlink points to nothing :
Redirect the softlink to the node object in the set of
data and continue.
The softlink points to the node object in its set of data :
All O.K. Continue.
The softlink points to a node object not in its set of data :
Have detected an illegal second registration. Fail.
When an attempt to add a Synonym softlink finds that the
softlink already exists. The application traverses the
softlink and finds what it points to.
The softlink points to nothing :
Redirect the softlink to the node object in the set of
data and continue.
The softlink points to the node object in its set of data :
All O.K. Continue.
The softlink points to a node object not in its set of data :
This Synonym name is already in use. Fail.
===============================================================================
Anyway,,,it sounds like it should do this BEFORE it attempts to execute
the DNS API CREATELINK call else DNS will check user access which
the user may not have...
Can this be adjusted in V1.2?
|
1760.2 | anybody out there? | FROSTY::JREGAN | | Mon Nov 11 1991 12:07 | 1 |
| anybody out there..???
|
1760.3 | anyone??? | FROSTY::JREGAN | | Wed Nov 13 1991 09:21 | 1 |
| anyone???
|
1760.4 | We're here, but coding comes before notes | TOOK::MINTZ | Erik Mintz | Wed Nov 13 1991 11:58 | 3 |
| I have forwarded you note to some people who may respond,
but it may be a while before any developer has time.
|
1760.5 | Please code the fix into V1.2 | CRONIC::LEMONS | And we thank you for your support. | Thu Nov 21 1991 15:41 | 20 |
| Hi!
I'm John Regan's field test site for DECmcc use of the corporate
namespace. I can certainly appreciate that coding comes before notes.
But John and I want to call the attention of DECmcc Engineering before
development of V1.2 is done. 1760.0 desribes a show-stopping bug in
the way REGISTER NODEV4 operates. Only by allowing unlimited access to
the synonym and backtranslation directories from everyone in the
network can this be worked around. Clearly, this is unacceptable.
I believe John has QAR'd this bug already. If DECmcc is used by a
customer with a glimmer of controlling access, they're going to be,
well, surprised.
Please let me know what I can do to help.
Thanks!
tl
|
1760.6 | I need an update on this. | CGHUB::JREGAN | | Fri Dec 06 1991 17:10 | 12 |
| I'm conferencing into a meeting in January with Westinghouse and
several Digital Sales Reps. It's a presales type meeting and they
want me to describe how Digital has designed, and implemented its
namespace. I plan to talk about naming, support, physical structure
and access control policies.
When I am asked if Digital uses DECmcc to manage node objects
named in the corporate namespace I'm going to have a problem
saying that we do,,,,because we don't and from what I can tell
we aren't going to.
What message should I be offering?
|
1760.7 | Changes will be made for V1.2 SSB | TRM::KWAK | | Tue Dec 17 1991 11:05 | 6 |
|
For DECmcc V1.2 SSB kit, we will make the changes in the softlink
creation routine. The MCC will check if the link already exists before
calling DECdns "dns$_create_link" function.
William
|
1760.8 | | CRONIC::LEMONS | And we thank you for your support. | Fri Dec 20 1991 14:21 | 4 |
| Great news! Thanks for doing this! When do you anticipate availability of this
fix?
tl
|