Skip to content

Latest commit

 

History

History
262 lines (138 loc) · 6.72 KB

GCP_Terraform_vars.md

File metadata and controls

262 lines (138 loc) · 6.72 KB
Raw

GCP Key Value Server Terraform vars documentation

  • add_missing_keys_v1

    Add missing keys v1.

  • backup_poll_frequency_secs

    Backup poll frequency for delta file notifier in seconds.

  • collector_dns_zone

    Google Cloud DNS zone name for collector.

  • collector_domain_name

    The domain name for metrics collector

  • collector_machine_type

    The machine type for metrcis collector

  • collector_service_name

    The metrics collector service name

  • collector_service_port

    The grpc port that receives traffic destined for the OpenTelemetry collector

  • collector_startup_script_path

    Relative path from main.tf to collector service startup script.

  • consented_debug_token

    Consented debug token to enable the otel collection of consented logs. Empty token means no-op and no logs will be collected for consented requests. The token in the request's consented debug configuration needs to match this debug token to make the server treat the request as consented.

  • cpu_utilization_percent

    CPU utilization percentage across an instance group required for autoscaler to add instances.

  • data_bucket_id

    Directory to watch for files.

  • data_loading_blob_prefix_allowlist

    A comma separated list of prefixes (i.e., directories) where data is loaded from.

  • data_loading_num_threads

    Number of parallel threads for reading and loading data files.

  • enable_consented_log

    Enable the logging of consented requests. If it is set to true, the consented debug token parameter value must not be an empty string.

  • enable_external_traffic

    Whether to serve external traffic. If disabled, only internal traffic via service mesh will be served.

  • environment

    Assigned environment name to group related resources. Also servers as gcp image tag.

  • envoy_port

    External load balancer will send traffic to this port. Envoy will forward traffic to kv_service_port. Must match envoy.yaml. Ignored if enable_external_traffic is false.

  • existing_service_mesh

    Existing service mesh. This would only be used if use_existing_service_mesh is true.

  • existing_vpc_id

    Existing vpc id. This would only be used if use_existing_vpc is true.

  • gcp_image_repo

    A URL to a docker image repo containing the key-value service.

  • gcp_image_tag

    Tag of the gcp docker image uploaded to the artifact registry.

  • instance_template_waits_for_instances

    True if terraform should wait for instances before returning from instance template application. False if faster apply is desired.

  • kv_service_port

    The grpc port that receives traffic destined for the frontend service.

  • logging_verbosity_level

    Logging verbosity level

  • machine_type

    Machine type for the key-value service. Must be compatible with confidential compute.

  • max_replicas_per_service_region

    Maximum amount of replicas per each service region (a single managed instance group).

  • metrics_export_interval_millis

    Export interval for metrics in milliseconds.

  • metrics_export_timeout_millis

    Export timeout for metrics in milliseconds.

  • min_replicas_per_service_region

    Minimum amount of replicas per each service region (a single managed instance group).

  • num_shards

    Total number of shards.

  • primary_coordinator_account_identity

    Account identity for the primary coordinator.

  • primary_coordinator_private_key_endpoint

    Primary coordinator private key endpoint.

  • primary_coordinator_region

    Primary coordinator region.

  • primary_key_service_cloud_function_url

    Primary workload identity pool provider.

  • primary_workload_identity_pool_provider

    Primary key service cloud function url.

  • project_id

    GCP project id.

  • public_key_endpoint

    Public key endpoint. Can only be overriden in non-prod mode.

  • realtime_updater_num_threads

    Amount of realtime updates threads locally.

  • regions

    Regions to deploy to.

  • regions_cidr_blocks

    A set of CIDR ranges for all specified regions. The number of blocks here should correspond to the number of regions.

  • regions_use_existing_nat

    Regions that use existing nat. No new nats will be created for regions specified here.

  • route_v1_to_v2

    Whether to route V1 requests through V2.

  • secondary_coordinator_account_identity

    Account identity for the secondary coordinator.

  • secondary_coordinator_private_key_endpoint

    Secondary coordinator private key endpoint.

  • secondary_coordinator_region

    Secondary coordinator region.

  • secondary_key_service_cloud_function_url

    Secondary key service cloud function url.

  • secondary_workload_identity_pool_provider

    Secondary workload identity pool provider.

  • server_dns_zone

    Dns zone for Kv-serer. Ignored if enable_external_traffic is false.

  • server_domain_ssl_certificate_id

    Ssl certificate id of the Kv-server domain. Ignored if enable_external_traffic is false.

  • server_url

    Kv-serer URL. Example: kv-server-environment.example.com. Ignored if enable_external_traffic is false.

  • service_account_email

    Email of the service account that be used by all instances.

  • service_mesh_address

    Service mesh address of the KV server.

  • tee_impersonate_service_accounts

    Tee can impersonate these service accounts. Necessary for coordinators.

  • telemetry_config

    Telemetry configuration to control whether metrics are raw or noised. Options are: mode: PROD(noised metrics), mode: EXPERIMENT(raw metrics), mode: COMPARE(both raw and noised metrics), mode: OFF(no metrics)

  • udf_num_workers

    Number of workers for UDF execution.

  • udf_update_timeout_millis

    UDF update timeout in milliseconds. Default is 30000.

  • use_confidential_space_debug_image

    If true, use the Confidential space debug image. Else use the prod image, which does not allow SSH. The images containing the service logic will run on top of this image and have their own prod and debug builds.

  • use_existing_service_mesh

    Whether to use existing service mesh.

  • use_existing_vpc

    Whether to use existing vpc network.

  • use_external_metrics_collector_endpoint

    Whether to use metrics collector on external host. On gcp, this should always be true because OpenTelemetry endpoints need to run on a different host

  • use_real_coordinators

    Use real coordinators.

  • vm_startup_delay_seconds

    The time it takes to get a service up and responding to heartbeats (in seconds).