HP OpenVMS Systems

ask the wizard
Content starts here

Mailbox SS$_NOREADER or SS$_NOWRITER timeout?

» close window

The Question is:

A server process owns a non permanent mailbox and is the only one reading to it.
Multiple processes are connecting to the same mailbox in write only mode.
When the server crashes, how long does it takes before the QIOs SENSEMODE
 READERCHECK issues by the clients return SS$_NOREADER.
In developpement stage I'm running the server interactively. When I <CTRL>Y and
 STOP et takes a few minutes before the clients receive the NOREADER status.
Seems like the mailbox is still alive and it takes a while before the OS is
 informed there are no more reader.

The Answer is :

  There is no particularly timeout interval defined -- that it takes
  as long as a few minutes would tend to imply the reader is busy and
  is not immediately honoring the message, or that all of the writers'
  connections are not being torn down as quicky as you appear to have
  In the present OpenVMS MBDRIVER implementation (and subject to change),
  the SS$_NOREADER and SS$_NOWRITER processing is called when a new I/O
  channel is assigned to the mailbox, or when a channel is deassigned.
  The OpenVMS Wizard would tend to use intra-cluster communications (ICC
  or a network transport and connections, and not mailboxes -- mailboxes
  tend to constrain application communications from being distributed
  across hosts, which can in term constrain the application deployment
  and scaling.

answer written or last revised on ( 30-DEC-2003 )

» close window