| Settings for experiments 115996669 |
| Started: Fri Dec 5 19:23:16 2025 -- up 1 hr 45 min 02 sec Built on Dec 5 2025 16:11:08 (1764979868) Built at rbex-enqueue-targets@oqs20.prod.google.com: Built as //cloud/ai/platform/dataplane/cardolan:vertex-genai-dataplane Build label: cloud-ml.vertex-genai-dataplane_20251205.04_p0 Built target blaze-out/k8-opt/bin/cloud/ai/platform/dataplane/cardolan/vertex-genai-dataplane Build options: fdo=XFDO Built for gcc-4.X.Y-crosstool-v18-llvm-grtev4-k8.k8 Built from changelist 840844128 with baseline 840844128 in a mint client based on //depot/google3 Task BNS: /bns/yudfwra/borg/yudfwra/bns/cloud-ml-vertex-genai-dataplane-staging-jobs/staging-qual-us.vertex-genai-dataplane/0 Ports: esf.0.staging-qual-us.vertex-genai-dataplane.cloud-ml-vertex-genai-dataplane-staging-jobs.yudfwra.borg.google.com:8112 0.staging-qual-us.vertex-genai-dataplane.cloud-ml-vertex-genai-dataplane-staging-jobs.yudfwra.borg.google.com:14013 borgenvelope.0.staging-qual-us.vertex-genai-dataplane.cloud-ml-vertex-genai-dataplane-staging-jobs.yudfwra.borg.google.com:14014 Profiling Links: censusprofilez?seconds=30 (CPU profile with go/census tags, 30 seconds): esf:8112 14013 borgenvelope:14014 censusheapz (heap usage with go/census tags): esf:8112 14013 borgenvelope:14014 peakheapz (peak heap usage): esf:8112 14013 borgenvelope:14014 deltacontentionz?seconds=10 (contention, 10 seconds): esf:8112 14013 borgenvelope:14014 threadz (thread stacks): esf:8112 14013 borgenvelope:14014 mmapz (mmap() usage): esf:8112 14013 borgenvelope:14014 contentionz (legacy contention): esf:8112 14013 borgenvelope:14014 Active RPC Experiments Stubby: stubby_default_subsetting, go/inline-psp, 1rpc, gRPC: channelz_use_v2_for_v1_api, channelz_use_v2_for_v1_service, channelz_zviz, chttp2_bound_write_size, deprecate_keep_grpc_initialized, error_flatten, event_engine_channelz_socket_info, event_engine_client, event_engine_dns, event_engine_dns_non_client_channel, event_engine_listener, event_engine_callback_cq, event_engine_secure_endpoint, google_no_envelope_resolver, graceful_external_connection_failure, lbns_support_in_address_resolver, loas2_protect_memory_optimization, max_inflight_pings_strict_limit, monitoring_experiment, namecheck_core_lib, privacy_context_single_encoding, prod2cloud_w3c_trace, rr_wrr_connect_from_random_index, tsi_frame_protector_without_locks, | Running on yudfwra-ap19.prod.google.com Platform: arcadia milan Process size: 13441MiB Memory usage: 1512MiB Load avg (1m): 37.81 View process information, endpoints View variables, flags, streamz, request logs Links: code, g3doc, continuous pprof, automon Distributed traces: view, change parameters Remote Logs: INFO WARNING ERROR STDOUT STDERR |
Name: V1pCountTokensRetries_Enabled_Launch::Experiment::ProdWave2 Experiment group: LAUNCH Owner email: jeremyafisher@google.com,mdb.vertex-prediction-team@google.com Aggregate experiment id: 115996663 Description: Prod partition for Wave2 mendel source: http://google3/googledata/experiments/vertex_ai/dataplane_prediction_common/studies/V1pCountTokensRetries_Enabled_Launch.gcl Query proportion: 0.00% Experiment layer: V1pCountTokensRetriesEnabledLaunchLayer Control experiment id: 115996655 Phase: NORMAL Diversion key: 2 Status: ACTIVE Managed traffic allocation: NOT_MANAGED End date: never Condition index: 0
V1pCountTokensRetries__retries_config: 'name: "V1pCountTokensRetries__retries_config"
type: PROTO_BINARY_BASE64
sub_type: "cloud_ai_platform_dataplane_prediction_proto.CardolanRetriesConfig"
base_value: "config_id: \"count_tokens_retries_config_experiment_002\"\nmodel_ids: \".*\"\nrequest_types: \"dedicated-critical_plus\"\nrequest_types: \"shared-critical\"\nrequest_types: \"shared-sheddable_plus\"\nerror_codes: DEADLINE_EXCEEDED\nerror_codes: RESOURCE_EXHAUSTED\nerror_codes: FAILED_PRECONDITION\nerror_codes: INTERNAL\nerror_codes: UNAVAILABLE\nerror_codes: INVALID_ARGUMENT\nretry_strategy {\n min_delay {\n nanos: 100000000\n }\n max_delay {\n seconds: 3\n }\n max_retries: 3\n request_deadline_fraction: 1\n}\nretry_threshold_fallback_behavior: RETRY_THRESHOLD_FALLBACK_BEHAVIOR_OPEN\n"
id: 0
'
V1pCountTokensRetries__enabled: 'name: "V1pCountTokensRetries__enabled"
type: BOOL
base_value: "TRUE"
id: 0
'