Conference azur::mcc

    I did not find in either the installation manual or the BMS Use manual
    any indication that RDB was required in order to start the Export
    Background Process.
    
    When I submitted the background process, there was no error indication
    in the resulting log file that said that RDB was not installed.
    
    I suggest that the error message surpression in the command file be
    moved untill after the environmental checks are complete and a specific
    warning message be written to the log file in the event that RDB is not
    found on the system.
    
    There are indications that RDB is required to use the suppled reporting
    package but not to simply do the exporting.
    
    s/rob

Rob,

I have a copy of the final version of the BMS use manual and the exporter 
chapter mentions the need for rdb all over the place starting right on
the first page of the chapter.

As for the Installation manual, you are right, the only place I could see
where it explained that RDB was required is on page 1-17, and this is specific
to the reports package.

jill

    I agree that the text "Rdb/VMS file" is used throughout. However,
    unless you KNOW that you need to install some software, you will be
    faced with at least a delay while your sales person scrambles to get it
    for you and since it is likely the system will need at least another
    reboot as a result of parameter changes for the RDB install, additional
    hassle is involved as well.
    
    What is missing is an explisit statement that you must install one of
    the RDB packages (and if more than run-time, the license too) in order
    to use either the EXPORT directive or start the background process.
    
    s/rob

I check the installation appendix on required stuff and it isn't there, that
alone isn't good since the reports section of the manual explicitly calls
it out as required.

    Rob , I've allready tried to find an answer of  this issue in note 643...
     Regards ,
    MaRc

    When attempting to DEREGISTER a NODE4, apparently the AM tries to
    contact the node. Why?
    
    In my case I am trying to remove a node that has moved to another
    DECnet area. The automatic update proceedures updated my DECnet
    database which is now out of sync with the info in DNS. I would like
    to remove the old entry and put back a clean one.
    
    For some reason when I attempt to deregister it I get the following
    error;
    
    DECmcc (V1.1.0)
    
    MCC> deregister node4 .host.nets
    
    Node4 CLAUDI_NS:.nssg.host.nets
    AT 18-MAR-1991 14:06:17
    
    The requested operation cannot be completed
                MCC Unhandled Service Reply = Access control information
    invalid at
                                              Node
    
    
    Even though the actual node can be accessed and there does not appear
    to be any node at the old address any longer.
    

    When specifying a SHOW EXPORT directive, the entity spec must be
    identical in case to the way it was registered. This is a pain if you
    don't remember how it was done. This only seems to be an issue for the
    Historian and Exporter. It should be case insensitive consitantly
    accross management modules.
    
    I got burned by an entity of the form ; :.LKG.host.took
    where I forgot that I had specified the location as upper case.
    
    In the Iconic Map It seems that even though you may have registered the
    entity with some mixed case, when you select it, and then do a
    directory, the management window shows the entity in all lower case.
    
    s/rob

    When I pull down the OPERATIONS menue and specify DIRECTORY, the
    management window comes up static. That is, it doesn't actually DO
    the directory but waits for you to push the button. This is not the
    same behaviour as you get with a SHOW.
    
    s/rob

    When requesting SHOW EXPORTING from the iconic map, and leaving the
    export target blank, the value of the logical name MCC_EXP_RDB_FILE ,
    described in the manual on page 4-12, is not used. Instead, it appears
    to use the SYS$LOGIN:MCC_EXPORT.RDB (or perhaps something else?).
    
    Also, it is not clear what the little red button is for next to the
    export target line. The manual does not describe it's use.
    
    s/rob

    Is ther some reason why I can't select text from management windows
    with the mouse and copy it? It only seems to work for "set" windows but
    would seem to be usefull for "show" or "directory" windows too.
    
    s/rob

Re .8

For all action directives, whether there are arguments or not, the user must
press the start operation before the operation commences.  See Zero, Enable,
Test,...  

If this is not a desired behavior, it can be changed but it is working as
designed and coded.

Verna

    Two points;
    
    First, for a SHOW operation the user does not have to push a button to
    start the operation. I think that is the desired behaviour.
    
    Second, what benefit to the customer did you (meaning engineering, not
    you specificly Verna) have in mind when it was designed to require an
    additional (but apparently unneeded to my mind) action on the user's
    part? The info would be helpful in the event of explaining this
    behaviour as a feature to a customer.
    
    Thanks
    
    s/rob

Rob,

It is hard to remember exactly why the decision was made BUT part of it was
to ensure that once a entity and action directive are selected, the user can
double check that their selections were correct (ie selected Enable not Disable).

For example, if the user fat-fingered and selected Zero Node4 and it was done
automatically, the results would not be too good.  Action directives can do
some damage.  This way the user must start the operation - another chance to
cancel out if desired.

Any non-action examine directive (could be others besides Show) starts 
automatically, but this does no damage if the selection was incorrect.

Hope this helps a bit
Verna

    For the examples you mentioned, it sure makes sense to get a
    confirmation. However, why would DIRECTORY need confirmation? It
    doesn't require any arguments in the management window so I would
    expect it behave like a SHOW.
    
    Thanks for the explanation.
    
    s/rob

    to expand on .7;
    
    The new requirement as spelled out in the release notes, to not allow
    mixed case in NODE4 Synonyms and also to require exact case matching
    when requesting only Historian and Exporter functions is a problem.
    
    For the NODE4 synonyms, this restriction places an unreasonable burden
    on people who may have populated all or parts of their phase 4 systems
    into DNS with the V1.0 product.
    
    As to having to specify exact case for only a subset of permitted
    directives is inconsitant and contrary to the message for our customers
    that we have an integrated solution.
    
    I believe that these MUST be corrected for the V1.1 product. These
    restrictions are too important to let pass.
    
    s/rob

    I am trying out the reports package and wanted to get a report on
    the data I had been exporting (thanks to Sam !) so in order to answer
    the questions on what date and time to start and finish I used DECmcc
    to SHOW EXPORTING to look up the valid times for my data.
    
    However, what I saw surprised me. It showed me that it has not
    apparently run the most recent poll even though it claims to be active.
    Here is what got displayed;
    
    MCC> show exp node4 nssg line una-0 exp
    targ=user$nets:[r_spence.mcc_recording]test_exp.rdb
    
    Node4 nssg Line una-0
    AT 20-MAR-1991 13:13:10
    
    Exporting parameters are:
                            Exporting state = ACTIVE,
                                State since = 19-MAR-1991 11:44:04.76,
                              Export period = 0 01:00:00.00,
                                 Begin time = 19-MAR-1991 11:43:42.71,
                                   End time = 25-MAY-2012 00:00:00.00,
                              Export target =
    "USER$NETS:[R_SPENCE.MCC_RECORDING]TEST_EXP.RDB",
                               Request time = 19-MAR-1991 11:43:42.71,
                               Requested by = "R_SPENCE",
               Time of last successful poll = "20-MAR-1991 11:44:29.35",
                 Number of successful polls = 25,
                   Time of last failed poll = "NONE",
                   Last poll failure reason = "N/A",
                     Number of failed polls = 0,
                           Last export time = "20-MAR-1991 11:44:29.35",
                Time of last export failure = "NONE",
                 Last export failure reason = "N/A",
                  Number of export failures = 0,
                              Sequence name = "EXPORT",
                    Initial sequence number = 0,
                    Current sequence number = 25
    MCC>
    
    Note that the time I looked is 13:13:10 and the export period is set
    to 1 hour but the last export was at 11:44:29.
    
    What happened to the 12:44 export?
    
    s/rob

RE a few of them...

On the management windows taking immediate action instead of requiring the
user to puch a button; I would like to try to restate the problem based upon
the examples that you have give.

What you would like to see is:

	If the user selected, from the operations menu, a directive that
	is non-destructive, like a DIRECTORY command, and there are no
	arguments required...then the action should be taken without the
	user having to enter any additional mouse/keyboard commands.

The reason I reworded this, and Rob correct me if this isn't what you were
talking about, is that immediate action on destructive commands like DELETE
will tend to upset the "fat fingered" managers and the users of the net they
manage (especially if you took *THEIR* node off the next work by mistake).

The problem with the current im plementation is that the DIRECTIVE-TYPE field
in the MSL tells us what "type" of directive we have only in terms of what
type of information does it work on (attribute, events, arguments). These
types are defined in the SRM as EVENT, EXAMINE, MODIFY, and ACTION directives.
What we would need would be another field that would tell us if an ACTION 
directive is destructive or not. With the other types we know that MODIFY is 
destructive (that is the nature of the directive-types definition) and that
the EXAMINE and EVENT are non-destructive (that is the nature of their
definition). If we had this information we could probably have the PMs
expand their implementation (no committments, just passing around ideas 
here) to check this field and decide if immediate -- no uadditional user 
action required -- can be taken.

jill

    More on .9, SHOW EXPORT
    
    It turns out that the MCC_EXP_RDB_FILE logical MUST be a Process
    Logical for this to work. I had it as a /SYSTEM/EXEC logical and
    it didn't work. This seems like a bug.
    
    s/rob 

    re .-2; Jill's comments.
    
    Yes, Jill, the restatement of my request is correct. From a user view,
    the fact that SHOW operates immediately and DIRECTORY requires an
    additional action seems inconsistant. Now that you explained the
    internals I understand why it works the way it does, not that that
    changes my view of how I expect it to work.
    
    Thanks.
    

Glad to hear that the process logical worked!

    I have not been able to get either of the CIRCUIT Graphs from the
    reports package to work. All required software is installed per
    instructions. I can get the NODE and LINE reports to work ok.
    
    I get the following error from DECgraph broadcast to all my terminals
    
    Job MCC_RPTS_NODE4_CIRCT_DAILY_GRAPH_DTR (queue SYS$BATCH, entry 186)
    terminated with error status
    %SYSTEM-F-FILNOTACC, file not accessed on channel
    
    
    The batch log error looks like
    
    $ datatrieve
    VAX DATATRIEVE V5.1
    Digital Query and Report System
    Type HELP for help
    
    :node4_circt_hourly_graph
     exit
    $ graph/load/nointeracive/monochrome= CIRCUIT_id00000103211133
    CIRCUIT_id0000010
    3211133 mcc_reports_files:mcc_rpts_lgraph
    
    
    
    %SYSTEM-FILNOTACC, file not accessed on channel
      R_SPENCE     job terminated at 21-MAR-1991 11:35:32.06
     
    
    Both graph reports fail in the same way.
    
    s/rob

    
    Re: .6
    
    Deregister looks for the node on the network?
    
    Yup, it does.  We overgeneralized some of our validation and we
    check the node4 on deregister to make sure that it isn't a cluster
    alias.  Pretty silly, considering you're most likely to Deregister
    something after it has been put in a dumpster.  This has been QAR'd,
    and we'll fix it for V1.2.  Sorry about that.
    
    Re: .8 (and others)
    
    Directory.  First time I did a Directory of a node4 from the iconic map
    I stared at it for a few minutes before I noticed it wanted me to do
    something.  Since I do all of this from a VS2000 I figured all I had to
    do was be patient and it took awhile to notice that it wanted me to
    toggle the old "Start Operation" button.
    
    In general, the split behavior between Action and Examine directives
    makes sense, but certainly not in this case.  I think we should either
    special-case the Directory directive, or make it into an Examine
    directive (although I think there was a reason we couldn't do that).
    
    Anyway, I wicked agree that that behavior was most unappreciated.
    
    We may want to generalize to "non-desctructive action directives" but
    that is a much more complicated discussion.
    
    Let's just fix Directory, don't you think it would be swell?
    
    Re: .15
    
    Node4 synonyms all exact casing?  Personal commentary:  "Yucky."
    
    FWIW: In the DNA4 AM, we have to up case all that stuff in order to get
    it into and out of the DECnet databases.  DECnet Phase 4 isn't a
    case-sensitive environment.  In fact, it sometimes gets upset if you
    try to remove lowercase stuff from one of the databases.
    
    I'm sure we got stuck in a bad place on this one, but it sure would be
    nice if we could fix it in the future.  DECnet Phase 4 is not case
    sensitive, it is too bad to make it so in a tool.
    
    -Jim Carey
    

    re: .14 .17 .19 .22
    
    To put it simply,  the DIRECTORY directive is screwed up!  From the
    users point of view it should behave as an EXAMINE directive.  This is
    where the problem is.  An EXAMINE directive works only on attributes. 
    The DIRECTORY directive does not.   If it did, you would see the
    following in the pulldown menu:
    
    	Directory -> char
    		     counters
    		     status
    		     etc.
    
    Believe it or not it is actually defined in the MS as an 
    EXAMINE directive!  I had to hardcode the directive type to 
    ACTION because internally it processed as an Action directive. 
    It was easier to hardcode the Iconic Map than to have everyone 
    change their MS files.  It is supposed to be changed to  an
    action directive in the v1.2 product.
    
    I will look into what we can do to automatically execute it
    so it behaves like the SHOW directive.
    
    /Jim
    

    I have been playing with the EXPORTER background program and have
    discovered a problem with it's handeling of ^C. I have been testing it
    by running it interactivly. It appears to handle ^C the same as expected
    for ^Y. That is, it doesn't run down the image and close things up
    cleanly. In fact, it appears that a CONTINUE works after ^C.
    If you do a ^C after the EXPORTER has been running for a while, then,
    the next image you try and run generates a
    
       %MCC-ALERT_TERMREQ, thread termination requested
    
    message.

    When using help from the FCL in line mode and you have succesfully
    found the help you needed and then hit return several times to get an
    MCC> prompt, the screen is erased so the help information you just
    found is lost. WHY?
    
    I reported this as a problem in the V1.0 EFT and it ain't fixed yet.
    
    Please fix for sure in next release!
    
    s/rob

    Rob,
    
    This item is still qared. Unluckily fixing it is not as easy as
    it sounds.  In figuring out what to do there we two requirements
    we were given. Make long help page between each screenful and
    don't clear screen at the end of help. We have found that fixin
    both was more work then we felt we had time for. This is still
    on the todo list though.
    
    jill
    

    Thanks Jill. In general, clearing the screen when there is no specific
    need to clear it is a bad idea. Another place where it is very
    frustrating is when you exit the menue for the reports routines.
    
    In FORMS mode, screen stuff makes sense. In LINE mode, SCREEN stuff
    like clearing it, doesn't make sense.
    
    Using help in line mode is tough enough given the upside down way one
    has to look things up (yes, I understand why it is that way) but it is
    simply maddening to have it erased just when you want to copy the
    example as a directive.
    
    s/rob

    There are two ways (that I have found) to create a domain from the
    iconic map. The first is to OPEN it from the FILE menuebar. The second
    is to attempt to place it from the toolbox on to a domain map.
    
    If I create it by attempting to open it from the FILE pulldown OPEN
    DOMAIN, I get a very nice (thank you) pop up that says it doesn't exist
    and would I like to create it. If I push YES, I get another pop up
    that has the domain name, the Owner ID and Directory fields filled in
    with what look like reasonable defaults. I like this.
    
    If I create it by attempting to add it as a domain in another domain
    from the TOOLBOX I get a LARGE pop up that has the same fields as
    previous, DOmain name, Owner ID and DIrectory, but the Owner ID and
    Directory fields are blank, no defaults are filled in. This is
    inconsistant and anoying. The popups above (in previous methode) are a
    much better interface and should be replicated here too.
    
    Also, the initial popup that says "the domain doesn't exist, would you
    like to create it" is much more clear than the message in the second
    example, "Entity   Domain .mydomain   is not created. Please fill in
    the form below." ^                  ^
    Besides the      ^ funny spacing    ^ , the message wording seems like
    an attempt to create the domain has failed.
    
    s/rob
    

The difference is because when you open a domain, the code knows about 
the domain fm, builds the form specifically for it's create arguments.
(this is true for New Domain also)

When you add it from the toolbox, a generic routine is called that knows
nothing about the domain global entity.  It just grabs the create args
for the entity selected and displays them in the form.  

Now I know the user doesn't care about these implementation details.  S/he
just wants a consistent user interface.  

I will check on what can be done to fill in the values for these fields but
can't guarantee it will be done in v1.2. 


A future version of DECmcc should allow the MM developer to enroll a 
user action routine for each request argument.  This routine should 
be called to load the field with a specific default and perform a semantic
field validation, in addition to the syntactic check the iconic map does.
Maybe in v2.0?

/Jim

> Also, it is not clear what the little red button is for next to the
> export target line. The manual does not describe it's use.

   The red button lets the user decide whether or not to send the argument 
   across the call interface. 

/Jim

    I started DECmcc from a DECterm and just now I asked DECmcc to QUIT
    from the FILE Pulldown and while it was shutting down I went to the
    DECterm window and typed an uparrow and return. I just needed to exit
    and restart to see if the saved settings I just changed look right to
    me after they are loaded. I got the following error that may be of
    interest.
    
    $ manage/ent/inter=decw
    $ manage/ent/inter=decw
    X Toolkit Warning: BadImplementation - server reported implementation
    error
    XIO:  fatal IO error 65535  on X server "NETS::0.0"
          after 154 requests (149 known processed) with 0 events remaining.
    %XLIB-IOERROR, xlib io error
    -DECW-CNXABORT, connection aborted
    
    I typed a ^C after a few minutes where no cpu time was accumulating
    and tried it again.
    %MCC-CANCEL, cancel
    $
    
    Got the error again and this time when I typed ^C I got a major dump
    
    31> manage/ent/inter=decw
    X Toolkit Warning: BadImplementation - server reported implementation
    error
 >  %MCC-CANCEL, cancel
    XIO:  fatal IO error 65535  on X server "NETS::0.0"
          after 428 requests (341 known processed) with 0 events remaining.
    %XLIB-IOERROR, xlib io error
    -DECW-CNXABORT, connection aborted
    %SYSTEM-ACCVIO, access violation, reason mask=04, virtual
    address=00000000, PC=0
    02F67F3, PSL=03C00005
    %MCC-LOG_EVENT, logging event type 00000400 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:51:08.54
    -MCC-RTN_ENTRY, routine (MCC_ONCE) called at PC = 000FD541
    %MCC-LOG_EVENT, logging event type 00000400 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:51:08.58
    -MCC-RTN_RETURN, routine (MCC_ONCE) returns with the following status:
    -MCC-NORMAL, normal successful completion
    
    The MCC_MAIN image is now using about 28%-35% of my system and
    pagefaulting like crazy (and has been for 5 minutes now. I am going to
    type a ^Y to see if I can stop it.
    
    Well, I typed ^Y and it has gone into a loop (it seems). Here is some
    of the output.
    
    %MCC-LOG_EVENT, logging event type 00000200 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:56:11.89
    -MCC-RTN_ENTRY, routine (VMSLOCK_RELEASE) called at PC = 000FE6A2
    %MCC-LOG_EVENT, logging event type 00000200 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:56:11.92
    -MCC-VMSLOCK_RELEASE, resource=....MCC_EVT_MEMORY             ,
    VMSLck=0E2400A8,
     RefCnt=1, CurMode=5, GrantMode=0, QMode=0, Is_Blk=0
    %MCC-LOG_EVENT, logging event type 00000200 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:56:11.95
    -MCC-RTN_RETURN, routine (VMSLOCK_RELEASE) returns with the following
    status:
    -MCC-NORMAL, normal successful completion
    %MCC-LOG_EVENT, logging event type 00000200 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:56:12.05
    -MCC-RTN_RETURN, routine (MCC_LOCK_RELEASE) returns with the following
    status:
    -MCC-NORMAL, normal successful completion
    %MCC-LOG_EVENT, logging event type 00000040 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:56:12.25
    -MCC-RTN_ENTRY, routine (MCC_CONTEXT_GET) called at PC = 000EB759
    %MCC-LOG_EVENT, logging event type 00000040 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:56:12.54
    -MCC-RTN_RETURN, routine (MCC_CONTEXT_GET) returns with the following
    status:
    -MCC-NORMAL, normal successful completion
    %MCC-LOG_EVENT, logging event type 00000200 reported by thread (name =
    Main_Thre
    ad, id = 00010000) at  4-APR-1991 15:56:12.59
    
    
    I don't know how at this time I am going to stop it. Stop process I
    guess.
    
    s/rob

    An update... It appears that somehow the DECwindows Server musta run
    out of some resource. We tried new processes for DECmcc to no avail.
    The server process was screwed up so bad we had to reboot cause the
    server restart wouldn't work.
    
    After the reboot things worked again.
    
    s/rob

   I modified the code so that when you do a DIRECTORY the operation 
is started automatically.  This will be in the V1.2 kit.

/Jim

    There is an inconsistancy in the lable that DECmcc places in the window
    for a domain and therefore what the icon in the icon box is labled.
    
    For example I have 3 domains. Lets call them test1, test2 and test3.
    The top level is test1. It contains test2. Test2 contains test3.
    
    Ok, I open test1 and get a window with an icon labled
    	"DECmcc Domain test1"
    
    I double click on the test2 icon and the after we go down a level, the
    icon lable changes to
    	"test2"
    
    I then double click on the test3 icon and we get a lable of
    	"test3"
    
    Then, using MB2 to "go up", we get
    	"DECmcc Domain test2"
    
    If one only clicks on domains and goes up and down the tree, going
    down makes the lable only be the domain name while going up prepends
    the DECmcc Domain text.
    
    If one double clicks on a node4 however, then the DECmcc Domain text is
    included for example for a node4 george in domain test3 we get
    	"DECmcc Domain test3 Node4  .DNA_Node.george"
    
    Personally I suggest that the "DECmcc Domain" text be dropped in all
    cases.
    
    s/rob

    In the MCC_RPTS_NODE4_CIRCT_DAILY_GRAPH_DTR.COM file, the file's name
    is documented to be
    	MCC_RPTS_NODE4_CIRCUIT_DAILY_GRAPH_DTR.COM
    
    Took a while to find it when troubleshooting a reports problem.
    
    Please QAR it.
    
    Thanks
    s/rob

    I think I found a problem with DISPLAY ALARMS in the IMPM.
    
    I generated an alarm, then pulled down DISPLAY ALARMS and got the
    expected window. I then forgot about it and looked at other things
    which put the DISPLAY ALARMS window behind so I REALLY forgot about it.
    
    I then CLEARED all Alarms and then I caused another alarm.
    
    After that alarm went off, I DISPLAYed ALARMS again.
    
    The problem is that now that I have two DISPLAY ALARMS windows up, I
    can't get rid of both of them. Pushing OK on either window will cause
    the newest one to go away but nothing other than exitting MCC will
    trar down the other one.
    
    This was repeatable.
    
    s/rob

Rob,
Do you have access to the QAR system?

Fred

Rob,

I am putting this here to update people on our hallway conversation.
Since this isn't the only wierd problem you are seeing, there maybe
somthing more at work here. Your mcc has been up and running, doing
alot from what you told me, for a long time. Check to make sure that
the strange problems you are seeing aren't due to memory limits being
hit or any such thing, and get back to us with the results.

jill

    Well, I looked at the system and I don't see anything that jumps out at
    me as an obvious problem. Maybe I missed it so the next page has the
    output from SHOW MEMORY.
    
    If you see something that I should fix, please let me know.
    
    
    $ show mem/pool/full
                  System Memory Resources on 22-MAY-1991 14:05:19.64
    
    Small Packet (SRP) Lookaside List            Packets       Bytes       Pages
        Current Total Size                          1066      102336        200
        Initial Size (SRPCOUNT)                     1024       98304        192
        Maximum Size (SRPCOUNTV)                    3072      294912        576
        Free Space                                    50        4800
        Space in Use                                1016       97536
        Packet Size/Upper Bound (SRPSIZE)                         96
        Lower Bound on Allocation                                 32
    
    I/O Request Packet (IRP) Lookaside List      Packets       Bytes      Pages
        Current Total Size                           651      114576        224
        Initial Size (IRPCOUNT)                      441       77616        152
        Maximum Size (IRPCOUNTV)                    1323      232848        455
        Free Space                                    42        7392
        Space in Use                                 609      107184
        Packet Size/Upper Bound (fixed)                          176
        Lower Bound on Allocation                                 97
    
    Large Packet (LRP) Lookaside List            Packets       Bytes      Pages
        Current Total Size                            32       52736        103
        Initial Size (LRPCOUNT)                       20       32960         65
        Maximum Size (LRPCOUNTV)                      80      131840        258
        Free Space                                    17       28016
        Space in Use                                  15       24720
        Packet Size/Upper Bound (LRPSIZE + 80)                  1648
        Lower Bound on Allocation                               1088
    
    Nonpaged Dynamic Memory
        Current Size (bytes)        999936    Current Total Size (pages)   1953
        Initial Size (NPAGEDYN)     999936    Initial Size (pages)         1953
        Maximum Size (NPAGEVIR)    1999872    Maximum Size (pages)         3906
        Free Space (bytes)          263760    Space in Use (bytes)       736176
        Size of Largest Block       217440    Size of Smallest Block         16
        Number of Free Blocks          124    Free Blocks LEQU 32 Bytes      31
    
    Paged Dynamic Memory
        Current Size (PAGEDYN)     1000448    Current Total Size (pages)   1954
        Free Space (bytes)          704304    Space in Use (bytes)       296144
        Size of Largest Block       701808    Size of Smallest Block         16
        Number of Free Blocks           82    Free Blocks LEQU 32 Bytes      76
    
    $ show mem
                  System Memory Resources on 22-MAY-1991 14:10:16.50
    
    Physical Memory Usage (pages):     Total        Free      In Use   Modified
      Main Memory (24.00Mb)            49152         491       47951        710
    
    Slot Usage (slots):                Total        Free    Resident    Swapped
      Process Entry Slots                 60          15          45          0
      Balance Set Slots                   45           2          43          0
    
    Fixed-Size Pool Areas (packets):   Total        Free      In Use       Size
      Small Packet (SRP) List           1066          44        1022         96
      I/O Request Packet (IRP) List      651          38         613        176
      Large Packet (LRP) List             32          17          15       1648
    
    Dynamic Memory Usage (bytes):      Total        Free      In Use    Largest
      Nonpaged Dynamic Memory         999936      262416      737520     217440
      Paged Dynamic Memory           1000448      703824      296624     701424
    
    Paging File Usage (pages):                      Free  Reservable      Total
      DISK$SYS_NETS:[SYS0.SYSEXE]PAGEFILE.SYS      31646        6868      50000
      USER$NETS:[SYS0.SYSEXE]PAGEFILE.SYS;3        71079       12661      99992
    
    Of the physical pages in use, 6531 pages are permanently allocated to VMS.
    

the sizes all lot reasonable. but maybe it is a virtual memory problem?
How many domains did you have open (including children domains if they
are dynamic under the root where you opened the map),and did you have
notification enabled? also give an average size for the domain memberships.

jill

    Ok.  The picture looks like this;
    
                                LKG
                                 ^
         lkg1-1  lkg1-2 ... lkg2-a      lkg2-1  ...   10 domains, one static
                               ^            ^
                               ^        lkg2-1-c1  lkg2-1-c2  .... 8 domains
                               ^                        ^
                               ^        (5 domains, 2 node4, 1 snmp, 1 node)
             15 dynamic domains, 1 bridge, and 27 node4s
    
    Notification is enabled. The alarm rule was against one of the node4
    entities in the lkg2-1-c2 domain.
    
    Exporting is also going on (and has been for over a month) against
    two node4 entities, one line and one circuit on each, every 1 hour
    from the lkg2-a domain.
    
    There are only ever 10 or so alarm rules active at any one time.
    
    System is VMS V5.3-2 with 24mb and 100k free blocks on each disk.
    
    Normally I have opened the top level, clicked down to lkg2-a, up again
    and down to lkg2-1-c2. All the other upper domains are empty (or mostly
    so).
    
    s/rob

If I read your hierarchy tree correctly you are saying it consists
of 39 domains (dynamics only). For each domain the iconic map will
keep a listing (in memory) of the entity information, and a thread
will be generated to do notification on each of these. I believe
our thread stack size on the NOTIFY commands are currently set at 80
pages each. Could you try coming into your hierarchy lower down (some 
where, where you have 10 or less domains within the hierarchy), and see
if you still get the strange results. I am wondering if you are hitting
virtual memory limits and if the "strange" behavior is due to the
pms/fms attempting to recover.

just guessing, can you tell. Also you listed a local and cluster page
file, is the 50K the one you are using, and what are the limits
on your account, and do you have any other process logged in (say running
alarms in another job)?

Rob,

Not sure if this is your problem but it is note worthy anyway for those
concerned about MCC performance. I have been often asked lately what I
can manage with my station and my reply is "I'M OUT OF MEMORY (32.0M) 
AND I'M ONLY HALFWAY TO PRODUCTION". Rob you also appear out of physical 
memory (491 free from 49152 available) and this at the very least will
cause performance degradation. Who knows, it could be causing other
problems. I saw various things happen when I got to this point so it may
be having an effect on your MCC.

something to look into,
brad...

    I tried a new instance of MCC and opened the lower level domain
    directly (lkg2-1-c2) and the problems with notify for event alarms and
    the stuck window problem still exist.
    
    re .-1
    I know I am getting tight on physical memory but the pagefault rate on
    the system isn't bad and after all, it COULD swap something if it needs
    the memory. I agree that we are going to need some realistic system
    sizing numbers soon. The minimums are not useful for anything with more
    than a few dozen entities.
    
    Jill, here are the stats for my process. I think I followed the
    guidelines. BTW, thanks for helping (thanks to all the folks upstairs
    who are so helpful).
    
    
    Maxjobs:         0  Fillm:       100  Bytlm:        40000
    Maxacctjobs:     0  Shrfillm:      0  Pbytlm:           0
    Maxdetach:       0  BIOlm:        36  JTquota:       4096
    Prclm:           2  DIOlm:        18  WSdef:          512
    Prio:            4  ASTlm:        48  WSquo:         2048
    Queprio:         0  TQElm:       128  WSextent:     20000
    CPU:        (none)  Enqlm:      2048  Pgflquo:      40000
    
    s/rob

    RE:.44
    Rob,
    
      Your PageFileQuota is only 40,000 -- This means that's all the
    virtual memory you are going to get.  You should definitely bump this
    up since you are using the Iconic Map so much -- make it as big as you
    can get away with.  Also check the SYSGEN parameter VIRTUALPAGECNT (or
    something close to that), it should be slightly less than your page file
    size (size of PAGEFILE.SYS).  Remember that the PageFileQuota MUST be
    less than the VIRTUALPAGECNT.  While you are at it, I would bump up the
    ASTLM to 100.
    
    -Matt.
    (Actually, I'm starting to think that quotas/resources are not your
    problem, but it couldn't hurt to tune your system for running MCC a
    little better.)

    Thanks Matt. I will make the mods for my account and try it.
    
    BTW, the Virtualpagecnt is 100,000 (and pagefile is 120,000) so that
    seems ok.
    
    s/rob
    

    Some exporter questions/problems...
    
    I have had a set of export directives executing once per hour for
    over a month. I have been recording info about the export activity
    with pencil and paper about twice a day during this experiment.
    Over this past weekend my system, like most other systems around here,
    went down for some unknown length of time. When I checked the export
    info this morning I saw that the sequence number was now at 741 (it is
    now at 746) but as of noon last thursday it was at 706. This would
    indicate that lots of polling periods were missed (since there should
    have been 144 polls taken during that time, not 40) however, the
    output from a SHOW EXPORT doesn't support this conclusion.
    
    Also, the output claims NONE as the time of the last export failure but
    has a time for the date of the last failed poll. How can it have a
    succesfull export with a failed poll?
    
    The reason for the last failed poll was
          Last poll failure reason = "failed to call ETP",
    
    What does this mean?
    
    Here is the output:
    MCC> show export node4 islan1
    
    Node4 islan1
    AT 29-MAY-1991 14:28:29
    
    Exporting parameters are:
                        Exporting state = ACTIVE,
                            State since = 11-APR-1991 18:40:00.22,
                          Export period = 0 01:00:00.00,
                             Begin time = 11-APR-1991 19:00:00.00,
                               End time = 25-MAY-2012 00:00:00.00,
                          Export target = "USER$NETS:[R_SPENCE.DATA]EXPORT.RDB",
                           Request time = 11-APR-1991 18:39:56.38,
                           Requested by = "R_SPENCE",
           Time of last successful poll = "29-MAY-1991 13:04:47.33",
             Number of successful polls = 731,
               Time of last failed poll = "28-MAY-1991 09:04:47.29",
               Last poll failure reason = "failed to call ETP",
                 Number of failed polls = 15,
                       Last export time = "29-MAY-1991 13:04:47.33",
            Time of last export failure = "NONE",
             Last export failure reason = "N/A",
              Number of export failures = 0,
                          Sequence name = "bb",
                Initial sequence number = 0,
                Current sequence number = 746
    
    
    s/rob

    I have this problem with exporting. I tried this morning to start up
    some additional exporter background processes which each have their own
    RDB databases. This seemed to work fine and the new processes and
    sub-processes exist and use compute time every so often, and they have
    created the expected 2 files for RDB and the 5 files for MCC.
    
    I then started some exporting with the following command;
    
    MCC> export node4 CLAUDI_NS:.NSSG.host.nssg Export period = 0:05:00, -
    MCC>_ Export target = user$nets:[r_spence.data]export_node4.rdb
    
    and got the expected Exporting Started message.
    
    I then tried to check it out with;
    
    MCC> show export NODE4 CLAUDI_NS:.NSSG.host.nssg
    
    Node4 CLAUDI_NS:.NSSG.host.nssg
    AT 29-MAY-1991 14:22:19
    
    No exporting for the specified entity
    MCC>
    
    I am puzzled as to what I did wrong.
    
    s/rob

    Rob,
    
    Sorry about the late reply.  I've been away on vacation for 2 weeks.
    
    Regarding the Display Alarm window not going away, have you gotten an
    answer on that?  I'll look into it sometime this week.  
    
    Normally, when a Display Alarms window is already displayed, the
    Operation/DisplayAlarm menu item is dimmed out.  It only becomes active
    again when the display box goes away.  It could be that you stumbled
    on a case where the menu item became active again before the display
    box went away, in which case, any box except for the last one cannot
    be cancelled.  
    
    Christine
    

    Well, after 30 minutes of looking, I give up.
    
    Where in HELP is the list of valid relational operators that can be
    used for Alarms Rules?
    
    BTW, having subtopics in HELP that have exactly only one sub-subtopic
    under them is really frustrating when you are trying to find something
    in a hurry (remember, HELP is for quick lookups when you are away from
    the manual set...).
    
    s/rob

    In the FCL help for ENTITY - ALARMS - RULE - CREATE - Severity, it
    would be good to add the list of Severity levels here.
    
    s/rob

>    Where in HELP is the list of valid relational operators that can be
>    used for Alarms Rules?

For mcc 0 alarm rules, you can find the list under:

entity alarms rule create expression syntax

For the comparison format of domain rules, which is the only way users can
create rules from the IMPM you can find the (different) list under:

tutorial creating_rules comparison

I will enter a suggestion QAR that the domain rule help be re-worked to
include this information under the entity topics as well as under the
tutorial topics.

>    In the FCL help for ENTITY - ALARMS - RULE - CREATE - Severity, it
>    would be good to add the list of Severity levels here.

I will enter a suggestion QAR.

    Thanks to you too Pete.
    

    Notification has just about died on my system. I don't know why.
    
    A month ago I was playing with events and at that time discovered
    that I didn't get visual notification in the iconic map. A system
    crash put a stop to the investigaiton at the time.
    
    Last week I was doing a demo and noticed that normap alarms were not
    notifying consistantly on the map. It would work for one domain but
    not another.
    
    I tried deleting the domain and creating a new one of the same name
    to make sure it was dynamic (how DO you tell on an existing map
    anyway?). However, I still don't get notification on the map even
    with the new domain.
    
    I have started and exitted (not stopped) DECmcc several times. That
    doesn't seem to have an effect.
    
    I tried enabling notification from the FCL but have not seen anything
    even though it was enabled at one time where I did get a map color
    change (so maybe I didn't do it right?).
    
    Here is my notify command;
    MCC> notify domain zko-lkg events=(any events),entity list = (node4 *)
    
    After notify was enabled and the alarm proceedure had broadcast an
    alarm to my terminal but notify had not, I looked at the status of
    the alarm;
    
    MCC> show mcc 0 alarms rule zko-lkg_incir all status, in domain zko-lkg
    
    MCC 0 ALARMS RULE zko-lkg_incir
    AT  2-JUL-1991 14:47:02 Status
    
    Examination of attributes shows:
                                      State = Enabled
                                   Substate = Running
                    Time of Last Evaluation =  2-JUL-1991 14:46:50.73
                  Result of Last Evaluation = True
                            Error Condition = "No counters in the counters
                                              attribute partition for this
    entity.
                                              "
    
    
    Anyway... I don't know where to look. ANy help would be welcome.
    
    s/rob

    
    Hi Rob,
    
    I think there were problems with visual notification if the ALARMS
    patch was not applied.  Did you use the patch?
    
    MCC> show mcc 0 alarms rule zko-lkg_incir all status, in domain zko-lkg
    
    >> MCC 0 ALARMS RULE zko-lkg_incir
    >> AT  2-JUL-1991 14:47:02 Status
    
    >> Examination of attributes shows:
    >>                                  State = Enabled
    >>                               Substate = Running
    >>                Time of Last Evaluation =  2-JUL-1991 14:46:50.73
    >>              Result of Last Evaluation = True
    >>                        Error Condition = "No counters in the counters
    >>                                          attribute partition for this
    >>                                           entity.
    
    Could you provide the rule expression?  Did you use a sub-entity
    wildcard?  Did you use an EXCEPTION HANDLER procedure -- could
    this have been the broadcast you saw?  I'm confused by the fact
    that the Evaluation = True, but the Error Condition is filled in.
    It looks like the rule was evaluated more than once:  the former
    had an error, and the latter evaluated to true -- what do you think
    about this?
    
    aud...
    
    

    Ok, here is the rule;
    
    CREATE  MCC 0 ALARMS RULE zko-lkg_incir -
      EXPRESSION=(NODE4 bblk01 circuit syn-0 Inbound Utilization > 60 , -
      AT START=(+00:00:30) EVERY 00:05:00, -
      FOR START (-0:00) DURATION 00:04:30 ), -
      PROCEDURE=COM$:MCC_UTILIZATION_BROADCAST_ALARM.COM, -
      EXCEPTION HANDLER=MCC_COMMON:MCC_ALARMS_BROADCAST_EXCEPTION.COM, -
      CATEGORY="Circuit Utilization", -
      perceived severity=Critical, -
      DESCRIPTION="Inbound Utilization > 60%", -
      PARAMETER="USER=SYSTEM", -
      IN DOMAIN = zko-lkg
    
    MCC> show mcc 0 alarms rule zko-lkg_incir all char, in domain zko-lkg
    
    MCC 0 ALARMS RULE zko-lkg_incir
    AT  2-JUL-1991 17:15:13 Characteristics
    
    Examination of attributes shows:
                                  Procedure =
                 USER$NETS:[R_SPENCE.COM]MCC_UTILIZATION_BROADCAST_ALARM.COM;2
                      Exception Handler =
                           USER$NETS:[MCC]MCC_ALARMS_BROADCAST_EXCEPTION.COM;3
                            Description = "Inbound Utilization > 60%"
                               Category = "Circuit Utilization"
                              Parameter = "USER=SYSTEM"
                             Expression = (NODE4 bblk01 circuit syn-0 Inbound
                                          Utilization > 60 ,   AT
                                          START=(+00:00:30) EVERY 00:05:00,
                                          FOR START (-0:00) DURATION 00:04:30 )
                     Perceived Severity = critical
    
    *** Note, I had to mung the format above a bit to put it here - it
    looked ok in MCC
    
    I didn't notice the exception handeler go off although it may have.
    Most of the broadcasts were clearly from my alarms proceedure (a
    modification of the supplied one that also broadcasts the value).
    
    This proceedure worked great for me for hours at a time over several
    weeks. I haven't changed the proceedure. Notification doesn't seem
    to see it.
    
    I have the alarms patch installed. I don't have the PA patch installed
    but I think it only applies to bridges.
    
    s/rob

    Rob,
    
    When you said you were playing with events, were you talking about
    alarm rules using events (occurs function)?  The V1.1 IMPM only
    supports notification on alarm rules.
    
    When you re-added your domain to double-check that is was a dynamic
    member, did you also check that any of it's parents were static?  As
    long as there is one static member in the hierarchy (where the root is
    the domain specified in the Open Domain function), you will not get
    notification from the static member on downwards.  Also, there isn't
    currently a way to find out if the domain member is static or dynamic.
    Talk to Jim Lemmon if this is really important to you.
    
    Christine
    

    When I said I was playing with events, yes, I meant using OCCURS in
    alarm rules (since that is only use of events in V1.1).
    
    I always take the default when creating domains (except when I really
    want a static one). In the past the default has always been dynamic.
    If there is no way to check what I have, then how can I proceed?
    
    Perhaps the domain type should be added as a characteristic attribute?
    
    s/rob
    

    
    You can check the type of the domain via FCL. If the domain zko-lkg
    is a member of another domain called ZKO for instance, then use the
    following command -
                                              
    MCC> show domain ZKO member zko-lkg all char
    
    
    Domain WHO_KNOWS_NS:.ZKO Member WHO_KNOWS_NS:.ZKO_LKG
    AT  4-JUL-1991 10:05:45 Characteristics
    
    Examination of attributes shows
                                Member Type = Dynamic
                              Member Entity = Domain WHO_KNOWS_NS:.ZKO_LKG
    
    Cheers,
    Louis

    Thanks lots for the workaround for finding domain type.
    
    I hope that in the future you will be able to do a
    
    MCC> Show domain zko-lkg all char
    
    and see it also.
    
    s/rob

    I thought I had seen this before but I can't find it. Sorry if it is a
    duplicate.
    
    I was setting up EXPORTING the other day and started up export of
    a node4 but forgot to specify an end time. The export started up ok.
    
    I then selected Modify Export from the Operations menue and modified
    the form to include a stop time and pushed the start button.
    
    We got a no exporting for this entity error even though a SHOW EXPORT
    from the iconic map would show it to us. Though we got the error,
    it still successfully made the change.
    
    
    Associated with this, but common to most modify (or SET where there is
    allready stuff set) windows, it isn't very clear what the effect of the
    pushbuttons (the red ones down the left side) are during these
    operations. It seems that it doesn't matter if a button is red or not
    for attributes that you have not typed anything into the form.
    
    ???
    
    s/rob
    

Rob,

The purpose of the buttons on the left side of a set form 

	if they are on (the way they are initially displayed) AND there
	is a value in the field the value will be encoded after the user
	presses Start Operation

	if the button is off, even if there is a value in the field, it
	will not be encoded after the user presses Start Operation

Hope that helps.
Verna

    Thanks.
    
    Some feedback...
    It has seemed to me when I have used MODIFY activities (a SET where
    some attributes are allready set is really a modify also), I always
    seemed to expect that the buttons would be by default off and I would
    select the few attributes I want to change. I believe this will be
    a common operation for selected referance attributes with a wildcard
    entity spec in the future (when wildcard entity specs work for this
    sort of thing).
    
    s/rob

Rob,

Thanks for the feedback.  Others have expressed similar feelings about starting
with all field in the "Do not Send" toggle position.  Changes such as that have
been deferred for the time being (V1.2) but hopefully will be re-examined 
after.

Verna