dashboard_link parameter cannot be set when using SLURMCluster from dask_jobqueue

I am trying to use the dashboard with SLURMCluster, but setting dashboard_link=‘127.0.0.1:8787’ does not do anything. My code is as follows:

from dask_jobqueue import SLURMCluster

cluster = SLURMCluster(cores=32,
                     memory="100GB",
#                      project='P48500028',
#                      queue='premium',
#                      interface='ib0',
                    dashboard_link = '127.0.0.1:8787',
                     walltime='00:30:00')

cluster.scale(2)  # Start 100 workers in 100 jobs that match the description above

from dask.distributed import Client
client = Client(cluster) 

In fact the output of cluster is just:

Client
Scheduler: tcp://128.2.204.115:32776 | Cluster Workers: 2 Cores: 64 Memory: 200.00 GB

Am I missing something here?

[lesteve]

Weird, can you make sure that bokeh is installed in your environment.

Here is a conda command that creates an environment with the latest versions of dask, distributed and dask-jobqueue:

conda create -n testenv -c conda-forge dask distributed dask-jobqueue -y 

Activating this environment:

conda activate testenv

I can then execute the snippet above without any problems:

from dask_jobqueue import SLURMCluster

cluster = SLURMCluster(cores=1, memory='1GB', dashboard_address=':12123')
print(cluster.dashboard_link)

Side-comment (not related to your error): unless you have a good reason to use python 3.5, I would advise to use python >= 3.6.

[lesteve]

Great to hear that! I am going to close this issue.

Agreed the error could be better or maybe there could be a warning earlier if bokeh is not installed to say that the dashboard is not going to be available …

Feel free to open an issue in dask/distributed about this!