There are two answers to this question, depending on how you are doing
scheduling.

If you take the default (Which is Schedmode=polling), then what happens is this:

1.  You start the scheduler task on the client.
2.  The scheduler task contacts the server.
3.  The server either tells him to come back in 12 hours (if the schedule window
is more than 12 hours away), or tells him when to come back to do the backup.
4.  In either event, the session between the client and server is broken, and it
is the responsibility of the client to come back at the "appointed time".  To do
so, the scheduler simply starts a count down timer process, and goes to sleep
until the time limit expires.
5.  When the limit expires, the client reestablishes a session with the server,
and either initiates the backup, or gets a new time assignment.

Please note, that the responsibility for the actual backup is placed with the
client here,  If you reboot the machine, or turn it off, the server doesn't go
looking for it.  All communications protocols are supported with this method.

If you take the optional schedmode=prompted method, then:

1.  You start the scheduler task on the client.
2.  The scheduler task contacts the server.
3.  The server takes note of the request, and the IP address that the request
came in on.  He then completes the session with the client.
4.  When the window for the appropriate schedule opens, the Server scheduler
prompter goes out to the IP address that he got in step 3, and tries to
establish the session.  If he can, then the backup takes place.  If he cannot,
he goes into a retry mode.  After several attempts, he will issue a warning
message to the log.

Please note that this method only works with TCP/IP, and so is not supported on
all platforms, or with any other protocol.  If you turn off the machine, the
scheduler prompter will continue to contact the client whenever the window
opens.  If you change addresses, the new address is picked up when you start the
scheduler the next time, and start step 1 again.  There is no table to maintain
it uses the last known address.  As far as I know, the only way to stop the
search is to start up a polling mode session.

Hope this clarifies things for you.

Jerry Lawson
jlawson AT thehartford DOT com

______________________________ Forward Header __________________________________
Subject: node identification
Author:  INTERNET.OWNERAD at SNADGATE
Date:    7/23/97 9:11 AM


Question about how the server knows who the client is, and how
to find it for a scheduled backup.

Our guess:
----------
The dsmc schedule daemon must be running on the client for the
The dsmc schedule daemon must be running on the client for the
 server to perform a scheudled backup. When the daemon starts
up it contacts the server at which time it gets info
regarding when the next scheduled backup is to take place.

At that time, some info must be passed to the server (maybe IP, maybe
ethernet address) which enables the server to "find" the client
when the scheduled backup is ready to kick off.

We figure this is the case because nodes not in the local server
host file _AND_ not in the DNS are backed up just fine.

Can anyone confirm this and/or fill in the details?

TIA,
Sean

--
    *-------------------------------------------------------------*
    *-------------------------------------------------------------*
    |      Sean Curran            +     Epsilon                   |
    |      UNIX Sys Admin         +     50 Cambridge St.          |
    |      scurran AT epsilon DOT com    +     Burlington, MA 01803      |
    |      800-225-1970 x6858     +     (F)617-229-0407           |
    *-------------------------------------------------------------*
       According to my calculations, the problem doesn't exist.
                                        -Bumpersticker