api /v1/node/<id>/allocations returns an array of Allocation objects,…

[fernandezvara]

… not AllocationIndexStub

Endpoint for node allocations returns full allocations information instead of stub ones.

Probably the real goal for the endpoint was on return simpler information, so the endpoint to fix is on the server side instead of the client.

Anytime having that information there helps and simplifies getting info, so probably it can be interesting to have a /api/node/<id>/allocations and a /api/node/<id>/allocations?full=true with extended information.

Endpoint asked with a curl:

curl http://localhost:4646/v1/node/node-4/allocations?pretty=true
[
    {
        "ID": "0fe63e23-e7bf-28ea-c4c3-3703c07e5a13",
        "EvalID": "65824dc6-6416-c600-5a60-d1c4eb7d70ee",
        "Name": "111.cache[3]",
        "NodeID": "node-4",
        "JobID": "111",
        "Job": {
            "Region": "global",
            "ID": "111",
            "Name": "111",
            "Type": "service",
            "Priority": 50,
            "AllAtOnce": false,
            "Datacenters": [
                "ams3"
            ],
            "Constraints": [ ...],
            "TaskGroups": [ ... ]

...
...
]

[diptanu]

I think the original goal for returning lesser information was that we want listing operations to be performant. Adding more data adds to serialization cost.

[fernandezvara]

Yes, I know what you telling @diptanu , but since the endpoint returns full information I think it makes sense to get using the API client the full allocation information.

Probably nomad must allow to return the full or lesser information using some parameter being the default the summarised one, since there is already a cost converting on the servers the data to JSON.

So /node//allocations should return the []AllocationIndexStub and ?full=true (or something like that) []Allocation. That way the server and client will de/serialize the same amount of information.

Getting a summarised information forces to make 1 + n API requests instead of just 1.

Another posibility can be adding a function that like AllocationsFull returning the []Allocation and leave the current as is. In our use case the ideal is make just 1 request per server and calculate the current allocation to know if we need to spin up another node.

[dadgar]

Thanks @fernandezvara. Could you update the documentation https://github.com/hashicorp/nomad/blob/master/website/source/docs/http/nodes.html.md.

[fernandezvara]

Documentation updated.

[dadgar]

@fernandezvara: Thanks!

[github-actions]

I'm going to lock this pull request because it has been closed for 120 days ⏳. This helps our maintainers find and focus on the active contributions.
If you have found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.