Aws-parallelcluster: Static MAC addresses on the master and compute nodes

[denysthegitmenace]

I am trying to use cfncluster for running scientific computations. The software I use is licensed to specific MAC addresses.

How can I give my master node and compute nodes static MAC addresses every time I relaunch the cluster? I know I could use ENI with set MAC address but then it could only be attached to a master node?

Regarding the compute nodes. Here is an extract from the documentation.

In the case of very large clusters, it is often sufficient if you send us the output of the $ADFBIN/dirac info command from the head node and 2 or 3 compute nodes. As most compute node names have the same beginning, we can then add them with a wild card (for example Linuxchem*).

Can we somehow program names and MAC addresses of compute nodes to start with a pre-set string?

P.S. I might be speaking a good deal of nonsense. Please excuse me. I am not knowledgeable withe the subject of networking.

I would greatly appreciate you help. Thank you.

[enrico-usai]

Hi @dennisthegitmenace ,
you could change the name by using custom bootstrap actions but I cannot confirm that everything will work correctly (it should be verified).

One idea could be to launch the cluster with maintain-initial-size = true and use the same _public_ subnet for both master and compute nodes. In this way the nodes will have a "static" Public IPs and you can get the ENIs .

If you stop the Compute nodes the Public IPs and the ENI will be released, whereas the Elastic IP of the Master will be retained.

Another idea is to allocate some Elastic IPs and attach them to the nodes when they are running.

[denysthegitmenace]

Thank you so much Enrico. That added few things on my "to try" list. Currently I am waiting for some bureaucracy to be over with and hopefully I will get to try all of the above very soon.

[enrico-usai]

Ok @dennisthegitmenace , I'm going to close the issue, feel free to re-open it if you have other problems or to let us know what the final solution will be.

We suggest to use the latest version of AWS ParallelCluster (pip install aws-parallelcluster) that is the AWS supported tool built on the CfnCluster project.