Prompt Proto Service HSC Helm values reference

Helm values reference table for the prompt-proto-service-hsc-gpu application.

Key

Type

Default

Description

prompt-proto-service.additionalVolumeMounts

list

[]

Kubernetes YAML configs for extra container volume(s). Any volumes required by other config options are automatically handled by the Helm chart.

prompt-proto-service.affinity

object

{}

Affinity rules for the prompt processing pods

prompt-proto-service.alerts.server

string

"usdf-alert-stream-dev.lsst.cloud:9094"

Server address for the alert stream

prompt-proto-service.alerts.topic

string

None, must be set

Topic name where alerts will be sent

prompt-proto-service.alerts.username

string

"kafka-admin"

Username for sending alerts to the alert stream

prompt-proto-service.apdb.config

string

None, must be set

URL to a serialized APDB configuration, or the “label:” prefix followed by the indexed name of such a config.

prompt-proto-service.cache.baseSize

int

3

The default number of datasets of each type to keep. The pipeline only needs one of most dataset types (one bias, one flat, etc.), so this is roughly the number of visits that fit in the cache.

prompt-proto-service.cache.patchesPerImage

int

4

A factor by which to multiply baseSize for templates and other patch-based datasets.

prompt-proto-service.cache.refcatsPerImage

int

4

A factor by which to multiply baseSize for refcat datasets.

prompt-proto-service.containerConcurrency

int

1

The number of Knative requests that can be handled simultaneously by one container

prompt-proto-service.image.pullPolicy

string

IfNotPresent in prod, Always in dev

Pull policy for the PP image

prompt-proto-service.image.repository

string

"ghcr.io/lsst-dm/prompt-service"

Image to use in the PP deployment

prompt-proto-service.image.tag

string

"latest"

Overrides the image tag whose default is the chart appVersion.

prompt-proto-service.imageNotifications.consumerOffsetReset

string

"latest"

Kafka consumer offset reset setting for image arrival notifications

prompt-proto-service.imageNotifications.imageTimeout

int

20

Timeout to wait after expected script completion for raw image arrival (seconds).

prompt-proto-service.imageNotifications.kafkaClusterAddress

string

None, must be set

Hostname and port of the Kafka provider

prompt-proto-service.imageNotifications.topic

string

None, must be set

Topic where raw image arrival notifications appear

prompt-proto-service.instrument.calibRepo

string

None, must be set

URI to the shared repo used for calibrations, templates, and pipeline outputs. If registry.centralRepoFile is set, this URI points to a local redirect instead of the central repo itself.

prompt-proto-service.instrument.name

string

"HSC"

The “short” name of the instrument

prompt-proto-service.instrument.pipelines.main

string

None, must be set

YAML-formatted config describing which pipeline(s) should be run for which visits’ raws. Fields are still in flux; see the source code for examples.

prompt-proto-service.instrument.pipelines.preprocessing

string

None, must be set

YAML-formatted config describing which pipeline(s) should be run before which visits’ raw arrival.

prompt-proto-service.instrument.preloadPadding

int

42

Number of arcseconds to pad the spatial region in preloading.

prompt-proto-service.instrument.skymap

string

"hsc_rings_v1"

Skymap to use with the instrument

prompt-proto-service.knative.cpuLimit

int

1

The maximum cpu cores for the full pod (see containerConcurrency).

prompt-proto-service.knative.cpuRequest

int

1

The cpu cores requested for the full pod (see containerConcurrency).

prompt-proto-service.knative.ephemeralStorageLimit

string

"5Gi"

The maximum storage space allowed for each container (mostly local Butler). This allocation is for the full pod (see containerConcurrency)

prompt-proto-service.knative.ephemeralStorageRequest

string

"5Gi"

The storage space reserved for each container (mostly local Butler). This allocation is for the full pod (see containerConcurrency)

prompt-proto-service.knative.extraTimeout

int

10

To acommodate scheduling problems, Knative waits for a request for twice worker.timeout. This parameter adds extra time to that minimum (seconds).

prompt-proto-service.knative.gpu

bool

true

GPUs enabled.

prompt-proto-service.knative.gpuRequest

int

1

The number of GPUs to request for the full pod (see containerConcurrency).

prompt-proto-service.knative.idleTimeout

int

0

Maximum time that a container can send nothing to Knative (seconds). This is only useful if the container runs async workers. If 0, idle timeout is ignored.

prompt-proto-service.knative.memoryLimit

string

"8Gi"

The maximum memory limit for the full pod (see containerConcurrency).

prompt-proto-service.knative.memoryRequest

string

"2Gi"

The minimum memory to request for the full pod (see containerConcurrency).

prompt-proto-service.knative.responseStartTimeout

int

0

Maximum time that a container can send nothing to Knative after initial submission (seconds). This is only useful if the container runs async workers. If 0, idle timeout is ignored.

prompt-proto-service.logLevel

string

log prompt_processing at DEBUG, other LSST code at INFO, and third-party code at WARNING.

Requested logging levels in the format of Middleware’s –log-level argument.

prompt-proto-service.podAnnotations

object

See the values.yaml file.

Annotations for the prompt-proto-service pod

prompt-proto-service.raw_microservice

string

""

The URI to a microservice that maps image metadata to a file location. If empty, Prompt Processing does not use a microservice.

prompt-proto-service.registry.centralRepoFile

bool

false

If set, this application’s Vault secret must contain a central_repo_file key containing a remote Butler configuration, and instrument.calibRepo is the local path where this file is mounted.

prompt-proto-service.s3.auth_env

bool

true

If set, get S3 credentials from this application’s Vault secret.

prompt-proto-service.s3.disableBucketValidation

int

0

Set this to disable validation of S3 bucket names, allowing Ceph multi-tenant colon-separated names to be used.

prompt-proto-service.s3.endpointUrl

string

None, must be set

S3 endpoint containing imageBucket

prompt-proto-service.s3.imageBucket

string

None, must be set

Bucket containing the incoming raw images

prompt-proto-service.sasquatch.auth_env

bool

true

If set, this application’s Vault secret must contain a sasquatch_token key containing the authentication token for sasquatch.endpointUrl. Leave unset to attempt anonymous access.

prompt-proto-service.sasquatch.endpointUrl

string

""

Url of the Sasquatch proxy server to upload metrics to. Leave blank to disable upload. This is a preliminary implementation of Sasquatch support, and this parameter may be deprecated if we instead support SasquatchDatastore in the future.

prompt-proto-service.sasquatch.namespace

string

"lsst.prompt"

Namespace in the Sasquatch system with which to associate metrics.

prompt-proto-service.tolerations

list

[{"effect":"NoSchedule","key":"nvidia.com/gpu"}]

Tolerations for the prompt processing pods

prompt-proto-service.worker.grace_period

int

45

When Knative shuts down a pod, the time its workers have to abort processing and save intermediate results (seconds).

prompt-proto-service.worker.restart

int

0

The number of requests to process before rebooting a worker. If 0, workers process requests indefinitely.

prompt-proto-service.worker.timeout

int

900

Maximum time that a worker can process a next_visit request (seconds).