1. Unable to create/delete slice, takes forever for creation and errors out

Unable to create/delete slice, takes forever for creation and errors out

Home Forums FABRIC General Questions and Discussion Unable to create/delete slice, takes forever for creation and errors out

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • #2970
    Ertza
    Participant

      Hi all,
      I have been trying to create a slice using JupyterHub with the following code:

      
      slice_name = 'ultima2'
      site = 'MAX'
      num_nodes = 6
      network_name='net'
      try:
      i = 0
      ifaces = []
      slice = fablib.new_slice(name=slice_name)
      while i < num_nodes:
      print(f'Creating Node{i}')
      node = slice.add_node(name=f'Node_{i}', site=site, image='default_ubuntu_20')
      node.add_component(model='GPU_RTX6000', name=f'gpu{i}')
      node.set_capacities(cores=16, ram=64, disk=100)
      ifaces.append(node.add_component(model='NIC_Basic', name=f'nic{i}').get_interfaces()[0])
      i += 1
      net = slice.add_l2network(name=network_name, interfaces=ifaces)
      print("Submitting slice request")
      slice.submit()
      except Exception as e:
      print(f"Exception: {e}")
      

      This errors out after some time as:
      Exception: Timeout 600 sec exceeded in Jupyter wait

      I have the following questions:

      1. How to create this slice as I often encounter this error and have to keep trying again with new names as this goes into a state where its not created and hence cannot even be deleted as well. What is the correct way to create it or solve this error once it happens?

      2. How to delete this slice which is still provisioning? Using the code it says cannot delete as slice not found and I did not see anything in my slices webpage to delete it, btw the slices pages shows the following for this slice:

      This slice is provisioning now. The page will automatically refresh in 23 seconds, or reload the page.
      Reset Layout Download JSON Download PNG

      3. How to make Jupyter wait longer than 600 sec.

      Thank you.

      #2986
      Paul Ruth
      Keymaster

        There is a article here that links to some FABlib documentation.   The sphinx documentation here shows the details of the FABlib API.  Specifically, the submit function is described here and has an optional argument to set the timeout your are looking for.

        Also, be careful because that slice will only work if all 6 RTK6000s at MAX are available.  You might want to check the availability of those GPUs to find a site that has enough available.

        Paul

        #2998
        Ertza
        Participant

          Awesome, thank you for guide and links.

        Viewing 3 posts - 1 through 3 (of 3 total)
        • You must be logged in to reply to this topic.