[core] Add stats for the gcs backend for telemetry.

[fishbone]

Signed-off-by: Yi Cheng [email protected]

Why are these changes needed?

To get better understanding of how GCS FT is used, adding this metrics.

Test:

cat /tmp/ray/session_latest/usage_stats.json
{"usage_stats": {"ray_version": "3.0.0.dev0", "python_version": "3.9.12", "schema_version": "0.1", "source": "OSS", "session_id": "70d3ecd3-5b16-40c3-9301-fd05404ea92a", "git_commit": "{{RAY_COMMIT_SHA}}", "os": "linux", "collect_timestamp_ms": 1660587366806, "session_start_timestamp_ms": 1660587351586, "cloud_provider": null, "min_workers": null, "max_workers": null, "head_node_instance_type": null, "worker_node_instance_types": null, "total_num_cpus": 16, "total_num_gpus": null, "total_memory_gb": 16.10752945020795, "total_object_store_memory_gb": 8.053764724172652, "library_usages": ["serve"], "total_success": 0, "total_failed": 13, "seq_number": 13, "extra_usage_tags": {"serve_api_version": "v1", "gcs_storage": "redis", "serve_num_deployments": "1"}, "total_num_nodes": 2, "total_num_running_jobs": 2}}

Related issue number

Checks

• I've signed off every commit(by using the -s flag, i.e., git commit -s) in this PR.
• I've run scripts/format.sh to lint the changes in this PR.
• I've included any doc changes needed for https://docs.ray.io/en/master/.
• I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/
• Testing Strategy
  • Unit tests
  • Release tests
  • This PR is not tested :(

[jjyao]

Could you add a simple test? Shouldn't be hard.

[jjyao]

@thomasdesr could you take a privacy review?

[fishbone]

Could you add a simple test? Shouldn't be hard.

Do you have some sample code for this?

[jjyao]

Make sure to get @thomasdesr 's approval as well.

[thomasdesr]

Is it possible to constrain the value's to our expected set? e.g. ("memory", "redis")

Its fine if not, just curious!

[fishbone]

@thomasdesr I think it's possible (like we can do some validation). but I feel we probably need an update of the telemetry and make it a framework layer feature. The scope seems out of this PR. cc @jjyao

[jjyao]

Yea, when we register the key, we can also register all the valid values. But agree that it's out of scope of this PR.

[thomasdesr]

👍 Would be good candidate for the next time someone wants to work on the framework!

Being able to know if the values are bounded vs unbounded would be a good thing to have :D