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

Conference 44.370::system_management

Title:	system management communications forum

Moderator:	CHEST::THOMPSON

Created:	Fri Mar 21 1986
Last Modified:	Thu Jul 08 1993
Last Successful Update:	Fri Jun 06 1997
Number of topics:	490
Total number of notes:	2018

466.0. "The ADGTST note" by CURRNT::DAW (Pizzazz-man) Fri Mar 27 1992 10:18

    Hi All,
    
    I thought I'd start an ADGTST note to keep you aware of what's
    happening with the cluster, as well as migration issues.
    
    ADGTST is now available from the Solent LAT and should by now, have
    been defined on PASTIT, CURRNT and FUTURS. Please connect/set host to
    ADGTST instead of the individual cluster nodes. This will enable
    load-balancing to occur and the performance of your session is likely
    to improve by doing this.  If you are setting host from another machine
    which doesn't know ADGTST, please ask your system manager to update the
    DECnet node database on that machine for you.
    
    ADGTST will be migrated to VMS V5.5 over the next few days. Details
    will be posted here and mailed via the normal channels when it is
    available again.
    
    Regards,
    
    Rob

T.R Title User Personal
Name Date Lines

466.1 ADGTST, VMS V5.5-1 Queueing CURRNT::DAW Pizzazz-man Tue Jun 16 1992 09:15 46

T.R	Title	User	Personal Name	Date	Lines
466.1	ADGTST, VMS V5.5-1 Queueing	CURRNT::DAW	Pizzazz-man	`Tue Jun 16 1992 09:15`	46
	ADGTST has been upgraded to VMS V5.5-1MA (effectively VMS V5.5-1). EASE testing part 2 (the second testing) is currently being carried out. If your application has not yet been tested, it must have been by the end of next week. Contact me for access details. Below is an addendum to the DECStep V3.0 impact document. If your application makes use of the queueing sub-system of VMS then please be aware ! The following addendum will be available from tomorrow as EASE""::july1992:[docs]IMPACT_JULY1992.ADDENDUM_A EASE - July 1992 Impact addendum A. 15-June-1992 V1.0 With the introduction of the new queue manager for VMS V5.5: 1. Queue Manager and Job Controller functions are different. 2. A Single Queue Manager Process can act as a Cluster-wide server, accessing the queue database for all processes in the cluster. 3. Each Job controller, User process and symbiont on each node communicates directly with the centralised queue manager through a shared IPC link. Prior to VMS V5.5 a queue manager ran on each node in the cluster as part of the Job Controller process. Each nodes Job Controller/queue manager accessed a distributed database to control queuing. User processes symbionts and batch jobs communicated with theit local queue manager. Pre VMS V5.5 several User Authorisation files and Queue databases could exist on a cluster, with VMS V5.5 this is not possible.... - The VMS Version 5.5 Upgrade and Installation Manual states " A cluster has just one queuing system (one set of database files) and there is no limit to the number of queues that can be defined." It then follows that the queuing system must now be homogeneous on VMS V5.5 systems. It is perceived that this restriction will have little or ( hopefully ) no effect on our internal IS infrastructure. Notes 1044 and 518 in the Conference VMSNOTES discuss these issues.

	ADGTST has been upgraded to VMS V5.5-1MA (effectively VMS V5.5-1).

	EASE testing part 2 (the second testing) is currently being carried
	out. If your application has not yet been tested, it must have been
	by the end of next week. Contact me for access details.

	Below is an addendum to the DECStep V3.0 impact document. If your
	application makes use of the queueing sub-system of VMS then please
	be aware !

The following addendum will be available from tomorrow as
EASE""::july1992:[docs]IMPACT_JULY1992.ADDENDUM_A

EASE  -  July 1992 Impact addendum A.
15-June-1992
V1.0

With the introduction of the new queue manager for VMS V5.5:

  1.  Queue Manager and Job Controller functions are different.
  2.  A Single Queue Manager Process can act as a Cluster-wide server,
      accessing the queue database for all processes in the cluster.
  3.  Each Job controller, User process  and symbiont on each node
      communicates directly with the centralised queue manager through
      a shared IPC link.

   Prior to VMS V5.5 a queue manager ran on each node in the cluster as
  part of the Job Controller process. Each nodes Job Controller/queue
  manager accessed a distributed database to control queuing. User processes
  symbionts and batch jobs communicated with theit local queue manager. Pre
  VMS V5.5 several User Authorisation files and Queue databases could exist
  on a cluster, with VMS V5.5 this is not possible.... -

   The VMS Version 5.5 Upgrade and Installation Manual states
    " A cluster has just one queuing system (one set of database files)
      and there is no limit to the number of queues that can be defined."

  It then follows that the queuing system must now be homogeneous on VMS
  V5.5 systems.

  It is perceived that this restriction will have little or ( hopefully )
  no effect on our internal IS infrastructure. Notes 1044 and 518 in the
  Conference VMSNOTES discuss these issues.