Configure Cloudy Cluster (beta)¶
A YAML cluster configuration file for a Cloudy Cluster resource manager on an HPC cluster looks like:
# /etc/ood/config/clusters.d/my_cluster.yml --- v2: metadata: title: "My Cluster" login: host: "my_cluster.my_center.edu" job: adapter: "ccq" image: "my-default-image" cloud: "gcp" scheduler: "my_scheduler" bin: "/path/to/other/CCQ" jobid_regex: "different job_id regex: (?<job_id>\\d+) " # bin_overrides: # ccqstat: "/usr/local/bin/ccqstat" # ccqdel: "" # ccqsub: ""
with the following configuration options:
- This is set to
- The default cloud image to use when launching jobs. There is no default.
- The cloud provider being used. Valid options are
aws. Defaults to
- The name of the scheduler being used. There is no default.
- The path to the CCQ client installation binaries. Defaults to
- The regular expression to extract the job id from the ccqstat output. Defaults to
job id is: (?<job_id>\\d+) you. You should only need this if the ccqstat output changes format. If you are required to reconfigure, you’ll need to extract the named group
job_idas the default does.
- Replacements/wrappers for CCQ’s job submission and control clients. Optional
Supports the following clients:
Prompted for input¶
You may see this error when you initially try to start a job.
The /opt/CloudyCluster/srv/CCQ/ccqsub command was prompted. You need to generate the certificate manually in a shell by running 'ccqstat' and entering your username/password
This is because CCQ libraries require a certificate to be generated to communicate with the
backend servers. To remediate you’ll simply have to login through a shell terminal and generate
a certificate. Do this by running the
ccqstat command and entering your username and password
when prompted. If you’re successful, the command will generate a
ccqCert.cert in your home
directory that subsequent invocations will use.
Note these certificates expire, so you may have to generate them every so often or specify a very distant expiry date when you do generate them.