WARNING: The running instructions described in this page are for development purposes only. Users interested in deploying, running and controlling O²/FLP software or their own software with AliECS should refer to the O²/FLP Suite instructions instead.
This part assumes you have already set up the Go environment, fetched the sources and built all AliECS Go components.
The recommended way to set up a Mesos cluster is by performing a complete deployment of the O²/FLP Suite with o2-flp-setup.
The AliECS core on the head node should be stopped (systemctl stop o2-aliecs-core) and your own AliECS core should be made to point to the head node.
Typically, it can be done by replacing the AliECS core binary on the head node with your own and restarting the o2-aliecs-core systemd service.
The following example flags assume a remote head node centosvmtest, the use of the default settings.yaml file, very verbose output, verbose workflow dumps on every workflow deployment, and the executor having been copied (scp) to /opt/o2control-executor on all controlled nodes:
--coreConfigurationUri
"file://$HOME/workspace/Control/hacking/settings.yaml"
--globalConfigurationUri
"consul://centosvmtest:8500"
--mesosUrl
http://centosvmtest:5050/api/v1/scheduler
--verbose
--veryVerbose
--executor
/opt/o2control-executor
--dumpWorkflowsSee Using coconut for instructions on the O² Control core command line interface.
The AliECS core runs as a systemd service in the O²/FLP cluster at Point 2.
There is a checker script that polls AliECS for its status (coconut env list).
- The checker script runs via cron (checkAliECScore available in GL) and makes 3 attempts with 10 seconds timeout.
- All failed attempts are recorded in the aliecs local file /tmp/checkAliECScore.out
- The ILG message is issued at the third consecutive failure.