[openroad-developer] OR Server

Thu May 29 18:58:59 PDT 2008

We've got dozens of these daemon type applications serving the organisation.
We've found the most flexible arrangement is to run them as a regular frame
in a terminal server session then disconnect the session.

The operations staff can reconnect to the session from anywhere on the
network, click on the required app to see the status or make changes to its
behaviour whilst the process is running.  The trace window is locked down
and hidden using a winAPI call and trace log is stored in a LOG directory
which can be accessed via a file share.
Each screen can also be accessed from the main application in a normal
installation so end users can run the screen in an interactive mode -
wrapped in normal user security.

Many of these screens need to run in an interactive terminal because they
deal with network and internet resources that are available to a fully
logged in user.

Paul

-----Original Message-----
From: openroad-developer-bounces at lists.ingres.com
[mailto:openroad-developer-bounces at lists.ingres.com]On Behalf Of Kim
Ginnerup
Sent: Friday, 30 May 2008 4:59 AM
To: David Tondreau; openroad-developer at lists.ingres.com
Subject: SV: [openroad-developer] OR Server

Hi Dan

If I understand what you want you a client that can sit in the background
doint what ever needed exactly as you do in your userframe now.
Lets call that your service.

We have done this several times.
Build an application that can run in a ghost frame.
Let this very small frame be timer controlled by the userevent as you have
done.
But let it connect to an OpenROAD Server for all the SQL processing and
business logic.
This will make your the sevice client very small, because you only need it
to wake up and invoke some methods in the appserver.

This should work like a charm.

Now you need a way to get the service client to run on a server as a real
service.
We have done this on windows using srvany.exe
With this program you can wrap your OpeROAD client service as an NT-service.
Now you can start and stop it as all other services.

If you wan a small OpenROAD client to control the service You can write a
small eclient.
This eclient connects to the same appserver as you service client.
This makes it possible to start and stop your service.
just make a system call with a net start | stop "your service"

You can also use the appserver to query the service to see if it is alive.

In your service client you add another userevent that on regular basis write
a timestamp in the Appserver
You eclient can query this stamp to see if it is alive.

This may seem complicated, but you should eb able to make a POC in less than
half a day.
The srvany can be a bit complicated to understand, but it is not that bad
after you have done it once.
use the "sc" command to insall and configure the service.

regards

Kim

________________________________________
Fra: openroad-developer-bounces at lists.ingres.com
[openroad-developer-bounces at lists.ingres.com] P&#229; vegne af David
Tondreau [david.tondreau at ingres.com]
Sendt: 29. maj 2008 19:47
Til: openroad-developer at lists.ingres.com
Emne: [openroad-developer] OR Server

Hello!

I’ve got a relatively simple OR application that is one piece of our
overall SOA interface architecture. It polls a DB table every .X sec to
see if there are any records to process, does some simple SQL if there
is work to be done then goes back to the .X sec wait. I currently manage
the wait loop by using the delay attribute of the SendUserEvent command.
I initially had it running as a ghost frame but converted it to a user
frame so I can stop and start it via a GUI as well as set some runtime
options on the fly for debugging purposes.

This process needs to be running exclusively (one instance per DB) and
we’d like for it to execute within the scope of a server and not an
active Windows user session. The functionality I’m looking for is to
enable different clients to inquire about as well as control the status
of the process (stop, start, is it running?) . I initially thought this
might be a candidate for the OR app server but after reading the
documentation it looks to me that the OR server is not geared toward
this type of scenario – it doesn’t seem that you can initiate a
perpetual looping process/service that has no active client session
attached to it. Having to use a simple client to do the initial start of
the process isn’t a problem. Our scenario is that you would start the
process somewhere then disconnect while it continues to execute until
told otherwise, possibly by a completely different client session.

We’re interested in the app server concept for a couple of reasons but
primarily because this will ultimately be a ‘back room’ process that
shouldn’t need an active Windows user session context to continue to
execute under and can respond to commands/inquiries from different
remote clients within the scope of one executing instance.

Is my conclusion about the app server correct? Using java/tomcat, etc is
an option for us but we’re trying to avoid that route if possible.
Likewise it needs to be DBMS agnostic.

Thanks in advance for any suggestions or ideas you may be able to offer,

Dan

_______________________________________________
openroad-developer mailing list
openroad-developer at lists.ingres.com
http://lists.ingres.com/mailman/listinfo/openroad-developer

_______________________________________________
openroad-developer mailing list
openroad-developer at lists.ingres.com
http://lists.ingres.com/mailman/listinfo/openroad-developer