[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference azur::mcc

Title:DECmcc user notes file. Does not replace IPMT.
Notice:Use IPMT for problems. Newsletter location in note 6187
Moderator:TAEC::BEROUD
Created:Mon Aug 21 1989
Last Modified:Wed Jun 04 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:6497
Total number of notes:27359

2287.0. "Query/Prob: MIR Namespace Use and Maint" by COOKIE::KITTELL (Richard - Architected Info Mgmt) Thu Feb 06 1992 13:17

T1.2.4 VMS

I'm trying the non-DNS namespace to see if it is a good way to run our 
development environment, and let us avoid messing about with the DEC:
DNS namespace.

1. I notice that once MCC_DNS_SELECTION is set to "MIR", all names created in 
   the namespace seem to be lowercase. Using the IMPM I created a domain,
   entering the name as TEST_DOMAIN in all caps. It showed up from both PMs
   as LOCAL_NS:.test_domain.

2. Is there a utility equivalent to dns$control, for examining and modifying
   entries in the MIR namespace?

3. While on the one hand the intent is for the scope of the MIR namespace to
   be one node or cluster, what if the disk with the MIR files is DFS-mounted
   elsewhere? DFS doesn't allow shared access so we wouldn't be able to have
   concurrent updaters. But once we've got things registered, could things
   run on the DFS-mounted disk in read-only mode and still look up objects
   in the namespace?

4. Do we still access the namespace via the mcc_dns routines?

5. Should there be any problem setting up a namespace local to, for instance,
   our test system? I would copy the MCC_DNS_ENT and ...ATT.DAT files into
   a test system local directory, then make MCC_SYSTEM a searchist to find
   the local copies first. Should that do it? 

   I've done just that, and while I can create domains and access them from
   both PMs, any attempt to show or register our global entities from either
   FM results in %ERF-W-NOMSG, Message number 00088A60.
T.RTitleUserPersonal
Name
DateLines
2287.1TOOK::GUERTINDon't fight fire with flamesFri Feb 07 1992 07:0464
    Richard, I'll try to answer some of the questions:

>> T1.2.4 VMS
>>
>> I'm trying the non-DNS namespace to see if it is a good way to run our
>> development environment, and let us avoid messing about with the DEC:
>> DNS namespace.
>>
>> 1. I notice that once MCC_DNS_SELECTION is set to "MIR", all names created in
>>    the namespace seem to be lowercase. Using the IMPM I created a domain,
>>    entering the name as TEST_DOMAIN in all caps. It showed up from both PMs
>>    as LOCAL_NS:.test_domain.
>>
      That's right.  The DNS Local MIR stores fullnames in lower case. The
      MIR routines cannot handle case-insensive fullnames.

>> 2. Is there a utility equivalent to dns$control, for examining and modifying
>>    entries in the MIR namespace?
>>
      We use a development tool which uses DAP-like syntax/commands for
      reading/writing/deleting MIR records.  However, I believe there are
      plans for a scaled down program to simply dump and delete DNS Local
      MIR namespace entries.  We know this functionality is needed.
      
>> 3. While on the one hand the intent is for the scope of the MIR namespace to
>>    be one node or cluster, what if the disk with the MIR files is DFS-mounted
>>    elsewhere? DFS doesn't allow shared access so we wouldn't be able to have
>>    concurrent updaters. But once we've got things registered, could things
>>    run on the DFS-mounted disk in read-only mode and still look up objects
>>    in the namespace?
>>
      
      This should work.  The DNS Local MIR routines open up the same files
      for both read-only and read-write access.  When a DNS routine is
      called which requires write access (CreateInstance, WriteAttribute,
      etc.), the read-write Repositories get used, but otherwise the
      read-only repositories get used.  If/when someone directly or
      indirectly calls an mcc_dns routine which require write access to a
      DFS disk, there will be an error message.

>> 4. Do we still access the namespace via the mcc_dns routines?
>>
      Yes!  Please do not call the MIR routines directly, the records are
      stored in a special format.

>> 5. Should there be any problem setting up a namespace local to, for instance,
>>    our test system? I would copy the MCC_DNS_ENT and ...ATT.DAT files into
>>    a test system local directory, then make MCC_SYSTEM a searchist to find
>>    the local copies first. Should that do it?
>>
      Sounds good to me.

>>    I've done just that, and while I can create domains and access them from
>>    both PMs, any attempt to show or register our global entities from either
>>    FM results in %ERF-W-NOMSG, Message number 00088A60.
>>
      
      Pretty funky stuff.  Looks like QAR material to me.  Could you give
      more detail?  are MCC_SPECIFIC and MCC_COMMON the "usual" disks?  I
      test using the same setup that you describe (I copy the mcc dns files
      to my private directory) and I have never had any problems.

-Matt.
      
2287.2will investigate the error furtherCOOKIE::KITTELLRichard - Architected Info MgmtFri Feb 07 1992 10:3914
RE: .1

Matt,

   Excellent info, thanks. Seeing that you're using the setup I describe in
   .5, but aren't seeing the problem I report in .6, I'll go through and
   checks for shorts between my headphones,  ;-)  and if the problem still
   persists file a QAR with tons of detail.

   One thing that might give me a clue, though, can you give me an idea
   in which context the %ERF-W-NOMSG, Message number 00088A60 error might
   be generated?

   Richard
2287.3namespace directory creation?COOKIE::KITTELLRichard - Architected Info MgmtFri Feb 07 1992 11:189
Matt,

Does the MIR namespace need to be initialized ala MCC_DNS_SETUP? I didn't
do that, and haven't found a procedure that seems to help.

Thanks,

Richard
2287.4No init required for local NSTOOK::MINTZErik Mintz, DECmcc DevelopmentFri Feb 07 1992 11:405
>Does the MIR namespace need to be initialized ala MCC_DNS_SETUP? I didn't
>do that, and haven't found a procedure that seems to help.

No, no initialization is required for a local NS.

2287.5AM Bug - Namespace works fineCOOKIE::KITTELLRichard - Architected Info MgmtFri Feb 07 1992 15:113
Got it. I found and fixed a bug in the AM. I use the IMPM with the MIR namespace
now.