ocpiserve — command-line tool for making local OpenCPI containers available for remote access
When an application is run, the OpenCPI framework finds "local"
containers - those that are part of or directly attached to
the local system - where workers based on artifacts can execute.
A local system can use the ocpiserve(1) tool to offer its
local containers to other network-based systems for use as
"remote" containers.
The ocpiserve command-line tool is part of OpenCPI’s
remote containers feature, which adds containers
available in other systems on the network to the set
of containers considered for application execution.
The ocpiserve tool runs as a container server, serving up
local containers as remotely accessible containers to network clients.
Remote clients use ocpirun options to find the systems on the
network that are running ocpiserve and list the available containers.
When started, ocpiserve discovers all local network interfaces and
prepares to be contacted by clients for all of them. It optionally (if the
--discoverable option is used) also
prepares to receive multicast queries from clients on all network
interfaces. The ocpiserve container server is a fully multi-homed
server, using all network interfaces for normal use (TCP) discovery (via the
User Datagram Protocol (UDP)).
The ocpiserve container server does not run, initiate or control
entire OpenCPI applications, so it requires fewer resources than
ocpirun or ACI executables.
Artifacts - binary executables compiled from OpenCPI workers -
are not required to reside on the system running
ocpiserve. Instead, they are downloaded on demand
to the ocpiserve container server from
the remote system on which the application is initiated
and controlled and then cached for possible re-use. The ocpiserve
container server maintains this cache in a directory
named artifacts unless the --directory option is used
to change it. The cache is normally maintained after ocpiserve
exits unless the --remove option is used to delete it.
Starting ocpiserve locally requires logging in to the console on
the system with a username and password. CTRL/C can be used to
exit ocpiserve locally. Alternatively,
the ocpiremote(1) tool can be used to start,
stop, and otherwise control an ocpiserve container server
from a remote client.
Options can occur anywhere in the command for the user’s convenience.
When invoked without any options, ocpiserve displays information
about all of its options.
In the descriptions below, a plus sign (+) after the option indicates that it can be specified more than once on the command line.
--addresses=<file>, -a <file>
ocpiserve should write all its TCP/IP
addresses. See the description of the OCPI_SERVER_ADDRESSES_FILE
in the OpenCPI Application Development Guide for details.
--directory= <directory-name>, -D <directory-name>
artifacts.
--discoverable, -d
ocpiserve to receive UDP multicast discovery
queries from clients on all network interfaces.
--help, -h
ocpiserve.
--log-level=<value>, -l <value>
--loopback, -L
--onlyloopback, -O
--port= <value>, -p <value>
ocpiserve ports.
The default is to use dynamically assigned port numbers.
--processors=<number-of-RCC-containers>, -n <number-of-RCC-containers>
--remove, -r
ocpiserve exits.
--verbose, -v
ocpiserve executes.
Start the ocpiserve container server, allowing it
to be discoverable via UDP multicast from client systems
and displaying status and progress information:
ocpiserve --verbose --discoverable
Start the ocpiserve container server and direct it
to write its TCP/IP addresses to the file named myserveraddrs:
ocpiserve --address=myserveraddrs
Start the ocpiserve container server and direct
it to use the TCP port 9233 as the TCP port:
ocpiserve --port=9233