Skip to main content

Jülich Monitoring and Control

Project description

Installation | Execution | Security | Plugins | Further documentation | Getting help | Acknowledgments |


JuMonC stands for "Jülich monitoring and control" and allows easier access to monitor running simulations and provides data using a REST-API.


JuMonC can be installed using pip:

pip install JuMonC

This also creates an entry point, allowing to start JuMonC with just the command JuMonC

Runtime dependencies

Optional runtime dependencies


JuMonC allows installation with extras for the optional dependencies. Some of them need installed software on the system, that is not part of the PyPi installation, so that still has to be provided. It still makes the setup easier. Currently the following options for the extras are avaiable:

  • PAPI: For use of PAPI
  • NVLM: For use of NVML
  • GPU: For use of GPU functionality, for now only adds the use of NVML
  • psutil: For use of psutil
  • SSL: Only needed when using adhoc certificates
  • Full: All of the above dependencies provided at once
  • Devel: Includes the ful ldependencies and installs all tools neeeded for CI as well, to allow local tests

These extras can be used like this:

pip install JuMonC[psutil,SSL]


JuMonC needs to be started parallel to your running simulation and needs one mpi process on each node your simulation is using. The actual REST-API will only be avaiable on node 0 and then communicate internaly using mpi.

It can be started by running srun -n <#nodes> --ntasks-per-node=1 JuMonC & Because we want to run JuMonC in parallel with other srun jobs, it is important to reduce the allocated resources (especially the implicit), so other jobs can start as well.

  • --mem-per-cpu 1000M : reduce memeory for JuMonC
  • -G0 : do not allocate GPUs for JuMonC
  • --exact : do not over allocate

Depending on your usecase --oversubscribe or --overlap may be usefull as well.

By default, this will start the REST-API on port 12121 on rank 0 of this allocation. The port can be changed by providing -p=<port>. Depending on the firewall rules of the system you are using, it might be necessary to use a ssh-tunnel to forward this endpoint of the API to somewhere else where you can access it.



Security is always a concern, and to block unauthorized access of this API there is a scope based token authorization included. This means each action requires authorization, with different tokens having a different scope of allowed actions. The tokens included by default and printed to stdout on startup are hierarchical, therefore every token includes the authorized functions of all lower tokens:

  1. see_links
  2. retrieve_data
  3. compute_data
  4. retrieve_simulation_data
  5. compute_simulation_data
  6. full

There are two options to prove your authorization, either cookie based by providing the token once in the login function (e.g. /login?token=12345678) or providing it by every API call as an additional parameter (e.g. /v1/network?token=12345678).


JuMonC offers two different modes to use encryption for accessing the REST-API. One uses adhoc generated certificates, the other one allows you to supply your own certificates.

  1. adhoc certificates:
  • depending on the used software to acces JuMonC you will get a warning or by default (e.g. Mozilla firefox) a complete blocking of the connection (e.g. curl) that you need to accept or override. This is due to the certificates not having any chain of trust to already trusted certificates.
  • you need to have pyopenssl avaiable to use adhoc certificates
  • used with --SSL_ENABLED
  1. Providing certificates
  • You can generate your own self-singed certificates using for example openssl req -x509 -newkey rsa:4096 -nodes -out cert.pem -keyout key.pem -days 365
    • This leads to the same warings as the adhoc certificates, but allows a manual control of the signed certificate
  • using this methods needs three arguments, one for enabeling SSL --SSL_ENABLED and one each for the certificate and the key --SSL_CERT=/path/to/cert.pem --SSL_KEY=/path/to/key.pem. The paths can either be absolute paths or relatic paths from the current working directory.


JuMonC's functionality can be extended by plugins. To see about the development of your own plugins visit the LogParser plugin as an example, or see the plugin development documentation.

Currently avaiable plugins for JuMonC:

Further documentation

Getting help

Open an issue


Horizon Europe JuMonC has been developed as part of the CoEC project, the CoEC project has received funding from the European Union’s Horizon 2020 research and innovation programm under grant agreement No 952181.

CoEC Logo Horizon Europe

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

JuMonC-0.9.1.tar.gz (42.1 kB view hashes)

Uploaded source

Built Distribution

JuMonC-0.9.1-py3-none-any.whl (65.4 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Fastly Fastly CDN Google Google Object Storage and Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page