Conference iosg::all-in-1

2653.0. "NODE1::"73=" Stopped" by BARNA::DSMAIL () Wed Apr 30 1997 14:45

Hi,

A customer has a VMS cluster with NODE1 and NODE2,
he ran ALL-IN-1V3.0 from NODE1 without problems
Now, he runs ALL-IN-1 from NODE2, but he has File Cabinet problems.
    
SM/MFC/MS shows		             Server Name:  NODE1::"73="
                                          Status:  Stopped
        	            DECnet object number:  73
        		      TCP/IP port number:  7373
        
also:

                               Index of Servers
 (Selections: 0  )                                          (New messages: 0  )

   No.   Node   Num TCP/IP Type   Process Name       Startup Queue

 > 1     NODE1   73  7373  LOCAL  NODE1$SRV73        A1V30$BATCH


OAFC$SERVER_STARTUP.LOG;79
                     30-APR-1997 08:07:27.05

    No reference to NODE2!!!
    
search OAFC$SERVER_STARTUP.LOG NODE1

$       @sys$startup:oafc$startup SYS$SYSDEVICE:[ALLIN1.DATA_SHARE]
NODE1$SERVER73.DAT;1 NODE1$SRV73
            /PROCESS_NAME        = "NODE1$SRV73" -
$       WRITE/ERROR=write_mbx_error write_mbx "SYS$SYSDEVICE:[ALLIN1.
DATA_SHARE]NODE1$SERVER73.DAT;1"

                configuration_file: SYS$SYSDEVICE:[ALLIN1.DATA_SHARE]
NODE1$SERVER73.DAT;1
                process_name:       NODE1$SRV73


SH SYS 
VAX/VMS V5.5-2  on node NODE2  30-APR-1997 12:49:28.77   Uptime  11 04:21:23
  Pid    Process Name    State  Pri      I/O       CPU       Page flts Ph.Mem

20800219 SYMBIONT_0001   HIB      6  1486754   0 00:55:21.42     38565    293
2080221C NODE1$SRV73     HIB      6      332   0 00:00:05.91      1758   1988
20800257 OA$SYMBIONT_008 HIB      6       21   0 00:00:00.77       200    108
20800258 SYMBIONT_0009   HIB      5    14801   0 00:17:32.01      2142   1354
20802CA2 ALLIN1          CUR  1   4      713   0 00:00:07.33      3778    509
20802AF6 OA$FCV          SUSP     3        5   0 00:00:00.07        66    135
20801EF7 TM_SERVER       LEF      3       26   0 00:00:00.65       381    801




Starting server with the STA option
	NODE1::"73=" submitted to A1V30$BATCH for startup
Nothing happens
	Batch queue A1V30$BATCH, idle, on NODE2::
  /BASE_PRIORITY=4 /JOB_LIMIT=1 /OWNER=[SISTEMAS,SYSTEM] /PROTECTION=(S:E,O:D,
  G:R,W:W)


OAFC$SERVER_STARTUP.LOG;80
                     30-APR-1997 12:08:40.18
Contains as last lines:


$ start_server:
$       @sys$startup:oafc$startup SYS$SYSDEVICE:[ALLIN1.DATA_SHARE]
NODE1$SERVER73.DAT;1 NODE1$SRV73
$!
$!      Copyright information and description at end.
$!
$! Default any parameters not passed to us.
$!
$       IF P1 .EQS. "" THEN P1 = "OA$DATA_SHARE:OAFC$SERVER_CONFIG.DAT"
$       IF P2 .EQS. "" THEN P2 = "A1FS"
$!
$! Check that the configuration file exists before trying to use it for the
$! server.
$!
$       if f$search( P1 ) .eqs. "" then goto No_config_file
$!
$! We'll be running under a different UIC, we'll need CMKRNL for that, but the
$! subprocess will also need a number of other privileges, and hence we need to
$! make sure that we've already got them, since we can't pass them to the
$! subprocess otherwise. Set a symbol to the list so it's easy to change here,
$! in the error message, and in the RUN command.
$!
$       Required_privs = -
          "CMKRNL,SYSLCK,SYSNAM,SYSPRV,PRMMBX,TMPMBX,EXQUOTA,NETMBX,DETACH"
$       IF .not. F$PRIVILEGE( Required_privs ) THEN GOTO priv_server
$!
$! Check a process is not already running with the requested process name
$!
$       if f$type( Ctx ) .eqs. "PROCESS_CONTEXT" then -
            Temp = f$context( "PROCESS",Ctx,"CANCEL")
$       Ctx = ""
$!
$! Select processes that have the name the caller chose (or we defaulted to)
$! for the subprocess, are detached, are owned by OAFC$SERVER and are on this
$! node (the latter is the default).
$!
$       Temp = f$context( "PROCESS",Ctx,"PRCNAM",P2,"EQL" )
$       Temp = f$context( "PROCESS",Ctx,"JOBTYPE","DETACHED","EQL" )
$       Temp = f$context( "PROCESS",Ctx,"UIC","[OAFC$SERVER]","EQL" )
$!
$! Pick up the process ID, which clears the context if the process doesn't
$! exist, otherwise we know a process of the specified name already exists.
$!
$       Temp = f$pid( Ctx )
$       if f$type( Ctx ) .eqs. "PROCESS_CONTEXT"
$       then
$           Temp = f$context( "PROCESS",Ctx,"CANCEL")
$           goto proc_name_error
$proc_name_error:
$       WRITE SYS$OUTPUT -
          "A server process already exists with the name NODE1$SRV73"
A server process already exists with the name NODE1$SRV73
$       EXIT 0
%NONAME-W-NOMSG, Message number 00000000
$
$       write sys$output "Server is being started:"
Server is being started:
$       write sys$output "              configuration_file: ",config_file
                configuration_file: SYS$SYSDEVICE:[ALLIN1.DATA_SHARE]
NODE1$SERVER73.DAT;1
$       write sys$output "              process_name:       ",process_name
                process_name:       NODE1$SRV73

$ exit:
$       write sys$output ""

$       write sys$output ""

$       write sys$output "******************************************", -
                                                "******************************"
************************************************************************
$       write sys$output ""

$       write sys$output ""

$
$       exit
  ALLIN1       job terminated at 30-APR-1997 12:08:42.40

  Accounting information:
  Buffered I/O count:              45         Peak working set size:     433
  Direct I/O count:                57         Peak page file size:      3012
  Page faults:                    493         Mounted volumes:             0
  Charged CPU time:           0 00:00:01.20   Elapsed time:     0 00:00:03.42





Anybody could help us?

Regards,

		Josep
                            

    
    JoseP,
    
    It would appear as though the FCS is running, but the Management UI is
    showing it to be Stopped. When you go to SM MFC MFS does it take a long
    time before saying "Stopped" (~30 seconds?). If this is the case then
    your server is in trouble and is not responding to connection requests.
    The quickest solution is to STOP/ID the NODE1::"73=" process and 
    restart it from the SM MFC MFS subsystem.
    
    I don't think having the extra node running ALL-IN-1 effected this.
    
    In the longterm, there are a few things you should consider. Firstly,
    ALL-IN-1 V3.0 is *very* old so you should consider upgrading to V3.1 or
    (better still) V3.2. There are also a number of ICF patches for V3.1
    which address FCS problems which you should install if you cannot
    upgrade to V3.2.
    
    If you really have to stay on ALL-IN-1 V3.0, then at least install the
    TLC - TeamLinks Connection Package (to V2.1 and then V2.5), as this
    will also update your servers to newer, better versions.
    
    Finally, be aware of VMS patches which exist to resolve problems that
    appear to be with the servers. There are already many notes in this
    conference about them...
    
    -Kevin.
    
    
    

        Do you have DECnet/OSI there by any chance? Do you have cluster
        alias enabled on NODE2 by any chance? There used to be case that
        having alias outgoing enabled on the FCS object (or session
        control application) caused this kind of problem.
        
        _veli

    
    Ah..yes!  Veli is indeed correct. I didn't realise that it's from NODE2
    that you can't see the server running on NODE1 - my apologies, I was
    reading your note too quickly.
    
    The alias enabled could potentially be the problem...
    
    Kevin.
    
    

    The customer has only DECnet Phase IV.
    
    Any verification could be done in that?
    
    Best Regards,
    
    		Josep

    
    Josep,
    
    What does :
    
    NCP> SHOW EXEC CHAR
    
    show ?
    
    Kevin.
    
    

        $SHOW SYSTEM is all that you can do.  The server indicating the
        wrong status on the management menu is a known problem for version
        3.0 and 3.1, and fixed in 3.2 according to my IPMT.
        
        Dave Zaniewski

    
    Dave,
    
    Which problem is that ? I know there's a problem when you have lots of
    servers running (Max DASL sessions), but in this case they only have
    one or two servers running.
    
    Just curious,
    Kevin.
    
    

        Hi Kevin,
        
        HPXQB2DDE
        C961101-5797
        And Martin Cook references FC_SERVER_ICOS #189.0
        
        Dave Zaniewski

    
    Yes, this is the client exceeding the maximum number of DASL
    connections in one session (or to that effect), which is usually
    reproducable only when you have multiple FCS and AIDAs and you elect to
    view the status on each one (I think 12 is the critical number).
    
    So, this might not be the same problem, but I would certainly suggest
    they upgrade first and take it from there...
    
    -Kevin.
    
    
    

    There is not cluster alias.
	Before upgrade, could a relink be done to check if helps?
    
    Best Regards,
    			Josep