starting scheduler

% systemd start orcmsched

starting aggregator/daemon

% systemd start orcmd

stopping scheduler

% systemd stop orcmsched

stopping aggregator/daemon

% systemd stop orcmd

starting scheduler

% service orcmsched start

starting aggregator/daemon

% service orcmd start

stopping scheduler

% service orcmsched stop

stopping aggregator/daemon

% service orcmd stop

Scheduler runtime settings can be changed in

/etc/sysconfig/orcmsched

Aggregator and Daemon runtime settings can be changed in

/etc/sysconfig/orcmd

If you relocated Sensys without using configure switches to specify locations you will need to set environment variables on all nodes. See

• https://github.com/intel-ctrlsys/sensys/wiki/2.1.04-Relocate-RPM-Install-Path

Start ‘orcmsched’ as root daemon on the SMS – system management server (or on a head node for small scale clusters)

% orcmsched
[SMS-linux:96034] Sun Aug  3 23:39:11 2014: Sensys SCHEDULER [[0,0],0] started

Start ‘orcmd’ as a root daemon on the aggregator nodes (on a small size cluster, aggregator can run on head node alongside scheduler).

% orcmd 2>&1 | tee orcmd-log.txt

You should see output lines that include both general system information, and other lines that include temperature info; e.g:

% orcmd
[AGGREGATOR-linux:96071] Sun Aug  3 23:40:00 2014: Sensys aggregator [[0,0],1] started
2014-08-03 23:40:05-0700,AGGREGATOR-linux,41.000000,39.000000,35.000000, ...
2014-08-03 23:40:10-0700,CN-linux-01,41.000000,40.000000,36.000000, ...

Start ‘orcmd’ as a root daemon on the compute nodes. On Sensys Compute Node 1:

% orcmd
[CN-linux-01:26355]
******************************
Mon Aug  4 00:35:04 2014: Sensys daemon [[0,0],2] started and connected to aggregator [[0,0],1]
My scheduler: [[0,0],0]
My parent: [[0,0],1]
******************************

On Sensys Compute Node 2:

% orcmd
[CN-linux-02:26355]
******************************
Mon Aug  4 00:35:04 2014: Sensys daemon [[0,0],3] started and connected to aggregator [[0,0],1]
My scheduler: [[0,0],0]
My parent: [[0,0],1]
******************************

On Sensys Compute Node 3:

% orcmd
[CN-linux-03:26355]
******************************
Mon Aug  4 00:35:04 2014: Sensys daemon [[0,0],3] started and connected to aggregator [[0,0],1]
My scheduler: [[0,0],0]
My parent: [[0,0],1]
******************************

The ssh client and sshd needs to be setup to pass in the Sensys environment variables.

Here is an example using pexec:

Open the file /etc/ssh/sshd_config and add:

AcceptEnv OPAL_PREFIX OPAL_LIBDIR LD_LIBRARY_PATH

Open the file /etc/ssh/ssh_config and add:

SendEnv OPAL_PREFIX OPAL_LIBDIR LD_LIBRARY_PATH

% pexec -Ppm 'node[01-32]' --scp '/etc/ssh/sshd_config' %host%:/etc/ssh/.
% pexec -Ppm 'node[01-32]' --ssh 'service sshd reload'
# Copy over the Sensys release
% pexec -Ppm 'node[01-32]' --rsync '/opt/open-rcm' %host%:/opt/.

# After starting Sensys on HN, start on the CNs
% pexec -Ppm 'node[01-32]' --ssh '/opt/open-rcm/bin/orcmd'