[Release] Add 2.9.1 release logs

[architkulkarni]

Why are these changes needed?

Adds performance logs for cfbf98c.

python compare_perf_metrics 2.9.0 2.9.1
REGRESSION 10.41%: single_client_tasks_and_get_batch (THROUGHPUT) regresses from 8.429852592930626 to 7.552591739788697 (10.41%) in 2.9.1/microbenchmark.json
REGRESSION 7.08%: placement_group_create/removal (THROUGHPUT) regresses from 845.7511547073977 to 785.8508797030252 (7.08%) in 2.9.1/microbenchmark.json
REGRESSION 7.03%: single_client_put_gigabytes (THROUGHPUT) regresses from 20.6372354079233 to 19.186501513119794 (7.03%) in 2.9.1/microbenchmark.json
REGRESSION 5.45%: single_client_get_object_containing_10k_refs (THROUGHPUT) regresses from 13.112230033151658 to 12.397164753383363 (5.45%) in 2.9.1/microbenchmark.json
REGRESSION 5.10%: multi_client_put_calls_Plasma_Store (THROUGHPUT) regresses from 12988.08350923366 to 12325.324692399277 (5.10%) in 2.9.1/microbenchmark.json
REGRESSION 4.72%: 1_1_async_actor_calls_async (THROUGHPUT) regresses from 3320.575179316923 to 3163.777669072775 (4.72%) in 2.9.1/microbenchmark.json
REGRESSION 4.45%: 1_1_async_actor_calls_with_args_async (THROUGHPUT) regresses from 2415.0979141352886 to 2307.5841596274245 (4.45%) in 2.9.1/microbenchmark.json
REGRESSION 3.93%: single_client_wait_1k_refs (THROUGHPUT) regresses from 5.424892169925369 to 5.211488336217317 (3.93%) in 2.9.1/microbenchmark.json
REGRESSION 3.41%: n_n_async_actor_calls_async (THROUGHPUT) regresses from 23089.526825423094 to 22301.861041866883 (3.41%) in 2.9.1/microbenchmark.json
REGRESSION 2.70%: 1_1_actor_calls_concurrent (THROUGHPUT) regresses from 5354.497412488227 to 5209.763216908848 (2.70%) in 2.9.1/microbenchmark.json
REGRESSION 2.05%: multi_client_tasks_async (THROUGHPUT) regresses from 24316.337428119852 to 23817.030993400018 (2.05%) in 2.9.1/microbenchmark.json
REGRESSION 1.99%: client__tasks_and_put_batch (THROUGHPUT) regresses from 11472.04637188305 to 11244.160608396189 (1.99%) in 2.9.1/microbenchmark.json
REGRESSION 1.83%: tasks_per_second (THROUGHPUT) regresses from 586.1652936197164 to 575.425069261052 (1.83%) in 2.9.1/benchmarks/many_tasks.json
REGRESSION 1.44%: single_client_tasks_sync (THROUGHPUT) regresses from 1009.4349525282154 to 994.8959357892109 (1.44%) in 2.9.1/microbenchmark.json
REGRESSION 1.24%: single_client_get_calls_Plasma_Store (THROUGHPUT) regresses from 10676.942537676008 to 10544.047488771861 (1.24%) in 2.9.1/microbenchmark.json
REGRESSION 0.84%: 1_1_actor_calls_async (THROUGHPUT) regresses from 8802.650097299196 to 8728.36563834848 (0.84%) in 2.9.1/microbenchmark.json
REGRESSION 0.55%: n_n_actor_calls_async (THROUGHPUT) regresses from 26694.138600078164 to 26547.749741220032 (0.55%) in 2.9.1/microbenchmark.json
REGRESSION 0.55%: tasks_per_second (THROUGHPUT) regresses from 328.5366336052011 to 326.73502105628876 (0.55%) in 2.9.1/benchmarks/many_nodes.json
REGRESSION 0.45%: client__1_1_actor_calls_sync (THROUGHPUT) regresses from 530.5597986550025 to 528.1938300742655 (0.45%) in 2.9.1/microbenchmark.json
REGRESSION 0.29%: client__1_1_actor_calls_async (THROUGHPUT) regresses from 1012.4837493368098 to 1009.5892523668448 (0.29%) in 2.9.1/microbenchmark.json
REGRESSION 0.07%: 1_1_actor_calls_sync (THROUGHPUT) regresses from 2075.2443816745968 to 2073.7767760812935 (0.07%) in 2.9.1/microbenchmark.json
REGRESSION 45.67%: dashboard_p50_latency_ms (LATENCY) regresses from 5.978 to 8.708 (45.67%) in 2.9.1/benchmarks/many_tasks.json
REGRESSION 34.93%: dashboard_p95_latency_ms (LATENCY) regresses from 2569.856 to 3467.574 (34.93%) in 2.9.1/benchmarks/many_actors.json
REGRESSION 19.26%: stage_0_time (LATENCY) regresses from 13.148497581481934 to 15.681537866592407 (19.26%) in 2.9.1/stress_tests/stress_test_many_tasks.json
REGRESSION 5.97%: 3000_returns_time (LATENCY) regresses from 6.094248331000003 to 6.457862126000009 (5.97%) in 2.9.1/scalability/single_node.json
REGRESSION 5.54%: dashboard_p50_latency_ms (LATENCY) regresses from 23.754 to 25.07 (5.54%) in 2.9.1/benchmarks/many_actors.json
REGRESSION 4.40%: avg_iteration_time (LATENCY) regresses from 1.4471865177154541 to 1.5108911991119385 (4.40%) in 2.9.1/stress_tests/stress_test_dead_actors.json
REGRESSION 0.40%: avg_pg_create_time_ms (LATENCY) regresses from 0.9053212447438167 to 0.9089756411408867 (0.40%) in 2.9.1/stress_tests/stress_test_placement_group.json
REGRESSION 0.25%: dashboard_p99_latency_ms (LATENCY) regresses from 10057.867 to 10083.034 (0.25%) in 2.9.1/benchmarks/many_tasks.json

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 added any new APIs to the API Reference. For example, if I added a
    method in Tune, I've added it in doc/source/tune/api/ under the
    corresponding .rst file.
• 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]

REGRESSION 10.41%: single_client_tasks_and_get_batch (THROUGHPUT) regresses from 8.429852592930626 to 7.552591739788697 (10.41%) in 2.9.1/microbenchmark.json

Noise

REGRESSION 7.08%: placement_group_create/removal (THROUGHPUT) regresses from 845.7511547073977 to 785.8508797030252 (7.08%) in 2.9.1/microbenchmark.json

Noise

REGRESSION 7.03%: single_client_put_gigabytes (THROUGHPUT) regresses from 20.6372354079233 to 19.186501513119794 (7.03%) in 2.9.1/microbenchmark.json

Noise

REGRESSION 19.26%: stage_0_time (LATENCY) regresses from 13.148497581481934 to 15.681537866592407 (19.26%) in 2.9.1/stress_tests/stress_test_many_tasks.json

Noise.