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

Conference iosg::all-in-1

Title:ALL-IN-1 (tm) Support Conference
Notice:Please spell ALL-IN-1 correctly - all CAPITALS!
Moderator:IOSG::PYECE
Created:Fri Jul 01 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2716
Total number of notes:12169

2615.0. "DECnet OSI Problem ?" by LISA2::SEPULVEDA (The Sunshine People) Wed Apr 09 1997 20:46

    Hi All,
    
    I have just installed ALL-IN-1 v3.1 (Portuguese) on an Alpha.
    VMS 7.*
    DECnet OSI
    
    During the installation I got an error messages when starting Local
    File Cabinet Server
    
    %Run-S-PROC_ID, identification of created process is 000000B1
    %OAFC-W-NETLOST, The network connection to the File Cabinet Server was
    lost
    
    There are 32 of these messages and after that
    %OA-I-LASTLINE, ERROR STARTING SERVER ::"73="
    
    Most of the IVPs passed except for the File Cab. Server and Aida
    Server.
    
    After the installation I started ALL-IN-1 and :
    
    a) option MFC then Manage Partitions and there was not partition
    defined. The field Partition as only ::
    
    	If I try to edit this partition I get an error
    	Part$Maint:#OAFC_Partition.%Open[#OAFC_Partition]" not found
    
    b) I am able to start the sender but I am not able to start the
    fetcher.
    
    I also noticed that sys$node was not defined.
    
    My questions are:
    
    1) Is this in fact a DECnet OSI (bad config.) problem ?
    2) Is it enough to do a $define sys$node sota1/exec ?
    3) What could possible be the problem ? I am able to set h to other Vax
    systems on the network.
    There is a recurring message from OPCOM saying:
    Event: too few servers Detected from: node DTSS
    N� Detected 0
    N� Required 1
    
    
    Thank You all very much for all your help
    Best Regards
    Ana
    
    
T.RTitleUserPersonal
Name
DateLines
2615.1Some answersIOSG::STANDAGEMy hovercraft is full of eelsThu Apr 10 1997 10:4868
    
    Ana,
    
    Lots of things to talk about here...
    
    >>a) option MFC then Manage Partitions and there was not partition
    >>d>efined. The field Partition as only ::
    
    >>	If I try to edit this partition I get an error
    >>	Part$Maint:#OAFC_Partition.%Open[#OAFC_Partition]" not found
    
    The Manage Partitions subsystem you shouldn't need to be going to, it's
    only useful if you're running DNS naming (I've yet to find anyone that
    is after 7 years...)
    
    
    >>b) I am able to start the sender but I am not able to start the
    >>fetcher.
    
    >>I also noticed that sys$node was not defined.
    
    This last comment is probably the major problem. ALL-IN-1 and many
    other products rely on the existence of the SYS$NODE, otherwise you
    will end up with problems similar to that which you describe. 
    
    
    >>My questions are:
    
    >>1) Is this in fact a DECnet OSI (bad config.) problem ?
    
    Not exactly, DECnet OSI doesn't require SYS$NODE but as I said, a lot
    of digital products use it, I sure GAP could reply with all the gory
    historical details if necessary :-)
    
    >>2) Is it enough to do a $define sys$node sota1/exec ?
    
    No!
    
    The logical must be in the form NODE::
    
    Once this is defined I suggest you recreate all your servers (AIDA,
    FCS, Sender and fetchers) - although I don't know if you actually 
    have anything in your partition datafile. You might need to reinstall
    the product, and even then the update installation will probably not
    put things to right (it might need a fresh/new installation). Try
    creating the srevers first,and we'll take it from there...
    
    >>3) What could possible be the problem ? I am able to set h to other Vax
    >>systems on the network.
    >>There is a recurring message from OPCOM saying:
    >>Event: too few servers Detected from: node DTSS
    >>N� Detected 0
    >>N� Required 1
    
    
    DTSS is the Distributed Time Service Synchronizer, it's part of DECNet
    OSI and is the mechanism by which all your systems consolidate their
    system clocks so they're consistant and 'correct'. This isn't part of
    ALL-IN-1, so you'll need to look into the OSI documentation for more on
    this...
    
    
    -Kevin.
    
    

    
    
2615.2About the dtssVELI::KORKKOVeli K�rkk� @FNO, 879-5512Thu Apr 10 1997 20:0918
        Re .0
        
        About the DTSS. Maybe there are not any DECdtss servers in that
        LAN. If this is the case, you can either easily configure your
        system to supply the time or maybe the easiest thing to is just
        to block the event completely, like
        
        $ mc ncl block event dispatcher * global filter -
        	((Node, Dtss), Too Few Servers Detected)
        
        Of course effect of this command would disappear at next reboot
        so he proper (permanent) procedre would be
        
        1) copy sys$manager:net$event_local.template  to 'same'.ncl
        2) then edit that file to contain mentioned "block ..."
        3) execute the file $mc ncl @sys$manager:net$event_local.ncl
        
        _veli
2615.3about SYS$NODEVELI::KORKKOVeli K�rkk� @FNO, 879-5512Thu Apr 10 1997 20:1624
        Further, about the SYS$NODE as well as SYS$NODE_FULLNAME
        
        The proper way to fix these is to execute 
        
        $ @sys$manager:net$configure advanced
        
        and choose option 2, i.e. "Change namespace/naming information"
        (or something like that...). Go through the stuff and voila,
        SYS$NODE as well as SYS$NODE_FULLNAME are defined properly.
        
        After this is done, check whether there exists (right now!, but
        not after next reboot) a file
        SYS$MANAGER:NET$STARTUP_RENAME.COM.
        
        Check that file and if this file appears to rename your node
        back to something like empty name, just delete this file (before
        rebooting).
        
        Like .1 said, proper value of SYS$NODE is VERY ESSENTIAL to many
        products. Two phase commit protocol (TP_SERVER), LAT, Infoserver
        Client are few which fail to start if SYS$NODE is not defined
        properly.
        
        _veli
2615.4Lack of SYS$NODE is a symtom of a bigger problemTAY2P1::HOWARDWhoever it takesThu Apr 10 1997 22:1016
>        After this is done, check whether there exists (right now!, but
>        not after next reboot) a file
>        SYS$MANAGER:NET$STARTUP_RENAME.COM.
    
    I think it's best to check this file whenever you find it.  I found
    that EWS$DEFINE_NODE creates one of these files to rename your server
    to the name of the latest EWS node you added!  The problem is that you
    don't look for the file, and it only gets executed when you restart
    DECnet (or maybe only when you reboot).  Its syntax is fairly obvious
    to look at, so if it isn't doing what you want, delete it. Getting the
    node back to its correct name is a pain.
    
    Note also that many procedures check the existence of SYS$NODE to see
    if DECnet is up.  So don't define it unitl DECnet is up.  
    
    Ben