Conference spezko::cluster

What is SCSCONNCNT?  Increase it to 50 or 60 and try again.

    Thanks, I will try that as soon as I can pry the system away from my
"customer".

Steve

    In addition to the the symptoms in .1 I also see, during "$show clus/con"
(add connections), two things:

	LOCAL_PROC_NAME		CON_STA
	---------------		-------
	SCS$DIR_LOOKUP		CON_SEN		! This keeps comming and going
	MSCP$DISK		OPEN		! This is steady

    I still havn't gotten a chance to try changing SCSCONN yet.

Steve

    I'm working on this same cluster.  I tried setting SCSCONNCNT to 60...
    It didn't help.
    
    Any other thoughts?
    
    Thanks.
    
    -	stp

   What (non-zero) disk allocation classes are in use, and on which nodes?
   Does a bootstrap over the NI controller complete?  Are there any errors
   logged?  Is there a version of CLUSTER_AUTHORIZE in SYS$SPECIFIC:[SYSEXE]?

      .0, Steve dis you find a solution ?

Is it a known problem ? If an IPMT is needed, what else information would be
useful ?

Regards, 

Maria. 


Configuration :
-------------


Alphaserver 1000A 5/400 et 2000 4/233, cluster SCSI running OpenVMS V7.1.


 ------
 |    | Station alpha 4/233
 |    |   root SYS10
 |    |     (METEOR)
 |    |
 ------
   |
   |                            ETHERNET
  ------------------------------------------------------------------------
   |                                                |
   |                                                |
   |                                                |
 ----------                                     _____________  Alphaserver
 |         | Alphaserver 1000a 5/400            |            | 2000 4/233
 |         |   root SYS1 (MATEMA)               |            |  root SYS0
 |         |                                    |            |  (METIS)
 |         | SCSI Port allo class 200           |            |
 |         |        ____                        |            |SCSI Port Allo
 |         |--------|  |$200$DKA0               |            |    Class 300
 |         |  SCSI  ----                        |            |   ____
 |         |                                    |            |___|  |$300$DKA0
 | KZPSA   |                                    |KZPSA|KZPSA |   ----
 ----------                                     -------------
   |SCSI Port                              SCSI Port |   |SCSI Port
   |allo class                             allo class|   |allo class
   |301                                    301       |   |302 _____
   |                                                 |   |----|TAPE|$302$MKA400
   |        SCSI                        SCSI         |   |    -----
   |__________________|------|_______________________|   |    _____
                      |      | -$301$DKA100(SYSTEM DISK) |----|TAPE|$302$MKA500
                      |      | _$301$DKA200                   -----
                      |      | _$301$DKA300
                      |      | _$301$DKA400
                      -------
                                              



! VMS$DEVICES.DAT
CLUSTER_CONFIG created 17-mar-1997 14:25:50

[Port MATEMA$PKA]
allocation class = 301

[Port MATEMA$PKB]
allocation class = 200
!
!
[Port METIS$PKA]
allocation class = 300

[Port METIS$PKB]
allocation class = 301

[Port METIS$PKC]
allocation class = 302


>>> B fl 0,0 ewa0

...

%VMScluster-E-NOT_SERVED, Configuration change, the system disk is no
longer served by node METIS
%VMScluster-I-REINIT_WAIT, Waiting for access to the system disk server
%VMScluster-I-REINIT_WAIT, Waiting for access to the system disk server
%VMScluster-I-REINIT_WAIT, Waiting for access to the system disk server


....


On the two boot nodes, $ show dev/service ==> all disks are available.
                                                                           
No errors logged.

Only one cluster_authorize.dat in sys$common:[sysexe]

Boot was also done with only one boot node and scsconncnt = 60.

:Is it a known problem ? If an IPMT is needed, what else information would be
:useful ?

   Please start the IPMT, with the information here...  If additional
   information is needed, it will be asked for.

:Network load complete.
:Host name: MOLD
:Host address: aa-00-04-00-63-30
:%VMScluster-I-BUSONLINE, LAN adapter is now running 08-00-2B-B4-18-80
:%VMScluster-I-VOLUNTEER, System disk service volunteered by node MOLD
:     AA-00-04-00-63-30
:%VMScluster-I-CREATECH, Creating channel to node MOLD     
:08-00-2B-B4-18-80   00-00-F8-4A-A0-10
:%VMScluster-I-OPENVC, Opening virtual circuit to node MOLD    
:%VMScluster-I-MSCPCONN, Connected to a MSCP server for the system disk,
: node MOLD    
:%VMScluster-E-NOT_SERVED, Configuration change, the system disk is no
:longer served by node MOLD     FF-7F-00-00-83-00
:%VMScluster-I-REINIT_WAIT, Waiting for access to the system disk server

   I have no idea if this is significant, but MOLD appears to have
   more than a few Ethernet addresses here...

   Can you bring the satellite and the boot host onto the same LAN
   segment (a private LAN segment is even better), and eliminate any
   weird network hardware that might be lurking?

re .8 HOFFMAN:

>   I have no idea if this is significant

    Don't underestimate yourself, turns out we have a "invalid connection"
(PHY Status LED is flashing amber) on one of our DECswitch 900 EF'S
that is almost certainly causing my problem (not yet verified).

re .6 MENICACCI

>%VMScluster-I-REINIT_WAIT, Waiting for access to the system disk server
>%VMScluster-I-REINIT_WAIT, Waiting for access to the system disk server
>%VMScluster-I-REINIT_WAIT, Waiting for access to the system disk server

    I do not get a steady stream of REINIT_WAIT's. I get a stream of
MSCPCONN => NOT_SERVED => REINIT_WAIT. So it does not sound like the
same problem to me. It looks to me like my satellite is repeatedly
discovering the boot node and then loosing it.

Steve

	
Engineering has found the solution.

On this cluster, Decnet MOP and lancp MOP were both up and running.

The solution was to stop Decnet Mop and use Lancp MOP only.

Satellite booted OK.