fix: do not filter slots for mixed-slot-type pools

[carolinaecalderon]

Ticket

CM-503

Description

For agent based deployments, if agents of different slot types are assigned to the same resource pool, change the "Compute Slots Allocated" label to "Unspecified Slots Allocated" label to clarify the mixed statusfor the user.
Additionally, add an error log message in zero slot or multi-slot-type cases.
Finally, if a resource pool's slot type is of type TYPE_UNSPECIFIED, do not filter out any agents from the slot progress bar count.

See new label for slots allocated for pools with multi-slot-type agents (pool1) or zero-slot agents (pool2)

Test Plan

See unit tests in agent, confirming that the slot type assigned in the resource summary is "zero" or "unspecified" for zero or multiple slot type agents. See screenshots of the changes in the webui in the test cluster

For release party, I really think this should get manually tested -- you must spin up your own aws Ubuntu devcluster (reach out to me for instructions) and configure devcluster to have 1 agent with all the gpus and 1 agent with artificial slots.

You can access my demo cluster webui at http://54.84.91.59:8080/. (Message me for the password) Resource pool pool1 has multiple slot type agents (CUDA agent1 and CPU agent2). The configuration for these agents is:

  - agent:            
      # Each agent stage should have a unique name for devcluster.    
           name: agent1
      pre:
        - sh: make -C agent build
      config_file:
        master_host: 127.0.0.1
        master_port: 8080
        slot_type: cuda
        resource_pool: pool1
        container_master_host: $DOCKER_LOCALHOST
        # Often dtrain clusters have multiple gpus per agent.
        # Each agent needs a unique agent_id.
        agent_id: agent1
        visible_gpus: 3
  - agent:
      # Each agent stage should have a unique name for devcluster.
      name: agent2
      pre:
        - sh: make -C agent build
      config_file:
        master_host: 127.0.0.1
        master_port: 8080
        resource_pool: pool1
        container_master_host: $DOCKER_LOCALHOST
        artificial_slots: 8
        # Each agent needs a unique agent_id.
        agent_id: agent2
  - agent:
      # Each agent stage should have a unique name for devcluster.
      name: agent3
      pre:
        - sh: make -C agent build
      config_file:
        master_host: 127.0.0.1
        master_port: 8080
        slot_type: cuda
        resource_pool: pool2
        container_master_host: $DOCKER_LOCALHOST
        # Often dtrain clusters have multiple gpus per agent.
        # Each agent needs a unique agent_id.
        agent_id: agent3
        visible_gpus: 0, 1, 2

Checklist

• Changes have been manually QA'd
• New features have been approved by the corresponding PM
• User-facing API changes have the "User-facing API Change" label
• Release notes have been added as a separate file under docs/release-notes/
  See Release Note for details.
• Licenses have been included for new code which was copied and/or modified from any external code

[netlify]

✅ Deploy Preview for determined-ui canceled.

Name	Link
🔨 Latest commit	e831f0e
🔍 Latest deploy log	https://app.netlify.com/sites/determined-ui/deploys/66e4c1d0d21c94000890f4cf

[codecov]

Codecov Report

Attention: Patch coverage is 75.00000% with 3 lines in your changes missing coverage. Please review.

Project coverage is 54.52%. Comparing base (a58ed7c) to head (e831f0e).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
webui/react/src/components/SlotAllocationBar.tsx	0.00%	3 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #9902   +/-   ##
=======================================
  Coverage   54.52%   54.52%           
=======================================
  Files        1252     1252           
  Lines      156551   156557    +6     
  Branches     3597     3600    +3     
=======================================
+ Hits        85356    85369   +13     
+ Misses      71063    71056    -7     
  Partials      132      132           

Flag	Coverage Δ
backend	45.14% <100.00%> (+0.01%)	⬆️
harness	72.75% <ø> (ø)
web	54.33% <62.50%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
master/internal/rm/agentrm/resource_pool.go	39.74% <100.00%> (ø)
master/internal/rm/agentrm/summaries.go	100.00% <100.00%> (+39.13%)	⬆️
webui/react/src/utils/cluster.ts	100.00% <100.00%> (ø)
webui/react/src/components/SlotAllocationBar.tsx	16.40% <0.00%> (ø)

... and 5 files with indirect coverage changes

[johnkim-det]

one note but otherwise web LGTM

[ShreyaLnuHpe]

LGTM!