File expiry

Actual event source metadata that were used to create the bag.

Unique identifier of the bag.

Recorded time range.

Union of first and last log times of any event source is always a subset of this.

The cluster the logs originated from.

Event source metadata as claimed by on-prem via registration.

Human-readable description of the contents of the bag.

If known, the name of the solution running when the bag was created.

The total number of log items from all event sources.

The total number of bytes from all event sources.

The total number of uploaded log items from all event sources.

The total number of uploaded bytes from all event sources.

Unidirectional transitions only.

Optional reason for UNCOMPLETABLE or FAILED.

Used for terminal statuses, if an incomplete bag (with missing log items or event sources) is available.

A globally unique identifier for this blob that is used as the filename of the object backing it (might be on local disk or a distributed filesystem in the cloud such as GCS).

The recommended way to generate a blob_id is to concatenate a random 64-bit integer with a directory and a file-extension. For example: const std::string blob_id = absl::StrCat( “my_blob_dir/”, DataLoggerClient::GenerateUid(), “_detection.jpg”); log_item.mutable_blob_payload->set_id(blob_id);

Readers should handle the case when the object backing this blob does not exist, e.g. when the data was just written or has been garbage collected.

The contents of the file. Will be unset when reading from structured storage. Should be cleared on the client side when sent to the DataLogger, except when logging the blob itself (i.e. in LogItem.blob_payload).

The cluster the blob originated from.

The organization ID prefixes the bucket that logs are uploaded into.

The event source of the blob, if any.

Used to further prefix the blob ID to prevent collisions between blobs from different event sources.

The matching topic subscription(s) if present

Requested time range.

The layer of context below executive_session_id. Identifies the execution of a plan.

The layer of context below executive_plan_id. Identifies the execution of a skill.

The top context layer. Represents all of the work done by the executive from when it is initialized until it is destroyed. This is currently used by the frontend to load meshes, which would be expensive for the executive to log for every sequence plan.

This layer allows to tie data to a specific action in a plan.

This field is only set if the skill 'skill_id' is executed internally by another skill, specified by 'parent_skill_id'.

The ICON session controlling the part which is sending the logs. Only used for logs coming from ICON. Check has_icon_session_id to distinguish between an unset session ID and an actual session with ID==0.

The ICON action which is controlling the part which is sending the logs. Only used for ICOn logs and only set if an action is active. Check has_icon_session_id to distinguish between an unset action ID and an actual action with ID==0.

Logging associated with operations to acquire a perception scene.

The visualization context this log item is associated with. Used for visualizing Intrinsic recordings.

Additional optional key-value metadata labels, can also be used to filter on log item fetch.

This is the only field considered for filtering in the logger service.

If you want to filter on other Context fields, remember to also add them to this map.

A time range used to select from the LogItems available in the TimescaleDB rolling buffer.

A human-readable description of the recording, to make it easier to identify in the frontend.

The event sources to include in the recording, as regex patterns.

The metadata of the newly-created recording.

The logging level of the log message.

The name of the component that is sending the log message.

The body of the log message.

The file that the log message was logged from. This helps identify the source of the log message.

The function that the log message was logged from. This helps identify the source of the log message.

The line number that the log message was logged from. This helps identify the source of the log message.

The timestamp of the log message.

If set, downsamples the candidates by time interval. Setting this samples once every N periods.

For example: - List: [1s, 2s, 3s, 4s, 5s, 6s, 7s, 8s, 9s, 10s] - sampling_interval_time: 2s - Downsampled Output: [1, 3, 5, 7, 9]

If set, downsamples the candidates by count. Setting this samples once every N samples.

For example: - List: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10] - sample_interval_count: 2 - Downsampled Output: [1, 3, 5, 7, 9]

Time range of the logged items.

Total number of log items in the event source.

Total number of bytes of all messages from this event source.

Total number of uploaded items (for creating detailed progress messages).

Total number of uploaded bytes (for creating detailed progress messages).

Unidirectional transitions only.

For UNCOMPLETABLE or FAILED.

Sanitized.

Type information for Intrinsic protos and ROS message types respectively. Used as hints for translation or verification.

Not in use yet.

Pre-existing registered bag.

Cloud logs.

If true, generates a signed url to download the bag with.

Signed url to download the bag with, if available. https://cloud.google.com/storage/docs/access-control/signed-urls

Till when the signed url be valid.

Limits for the number of results per page.

Defaults to 10000.

Cumulative, defaults to 25MB

Session is bound to organization_id.

Set to true if blob payloads are to be omitted from the response. Note that blob_ids will still be populated regardless of this setting.

Generally used for pagination.

Opaque cursor to the next page of results.

If empty, no next page exists. Expires at next_page_cursor_expiry.

Metadata for this page of results.

Maximum number of items to return for this response. If unset, defaults to 10000.

The server may return fewer than this number of items even if there are more matching items available due to global throttles.

Returned in GetLogItemsResponse to continue where the previous call left off. Opaque to the client.

Best effort. If the data being requested has already been garbage collected, returns items starting at the beginning of the available data.

The LogItems that matched the criteria. This can sometimes be empty.

Opaque cursor to the next page of results. If empty, no next page exists at time of query.

This could be because, at time of query, either: - The next log item that could have been returned had a timestamp that was past the requested end time. - No log items were found.

After receiving a response with no next page cursor, if the client wants to continue reading (e.g., to fetch LogItems that were logged after the response was received), they must start a new GetLogItemsRequest.

If set, the response was truncated. There usually will be a next page cursor to continue reading.

Look up the log options for the given event source. Will return the first matching log options in the precedence order.

Look up the log options for the given key.

The key is just used to identify the log options for setting/overwriting and fetching, and is not used for matching against event sources.

The event_source to return the most recent LogItem for.

The most recently logged LogItem for the event_source.

A description of the topics

Requested time range.

Limits for the number of results per page.

Defaults to 1000.

Opaque cursor to the next page of results. If empty, no next page exists.

Optionally allow users to specify a starting bound for the recording list.

Optionally allow users to specify an ending bound for the recording list.

Allow users to specify whether to return a summary of the recordings or the full metadata, which includes statistics for each event source.

Optionally allow users to retrieve specific Bag IDs If no Bag ID's are provided, all bags will be returned that are in the optional time window.

Limits for the number of results per page.

Defaults to 1000.

Used for pagination.

Opaque cursor to the next page of results. If empty, no next page exists.

Limits for the number of results per page.

Defaults to 1000.

Used for pagination.

Opaque cursor to the next page of results. If empty, no next page exists.

Limits for the number of results per page.

Defaults to 1000.

Used for pagination.

Opaque cursor to the next page of results. If empty, no next page exists.

Truncate load if any of these parameters are met.

Defaults to 100000.

Cumulative, defaults to 1GB

Session is bound to organization ID.

Cache token for session.

When the logs will be deleted.

True if truncated (from client or server-side limits).

Opaque cursor to the first log in range, spanning the

Type-agnostic information about this event.

Type-agnostic information about this event that needs to be propagated throughout the gRPC call stack. This can be used to join this event with others that belong to the same timeframe.

Any structured data associated with this event.

Use this to log individual files smaller than 2GB (size limit for proto messages). For larger files sync them to GCS using the GCS API directly. If you have many files, consider creating a tar archive out of them first.

The event source to set the log options for. This must be a valid RE2 regex string, and will be used as such. Note there is an edge case where "." will match anything, so handle it accordingly.

For example: - "foo" will match "foo". - "." will match all event sources. - "foo|bar" will match "foo" and "bar". - "foo." will match "foo" and "foobar". - "foo|bar|baz" will match "foo", "bar", and "baz".

If set to 'true', logs are synced to the cloud.

The maximum byte size of the on-prem buffer for LogItems.

The token bucket options for rate limiting on the Log() rpc of the data logger.

Priority denotes value of items in the event_source.

Higher priority event sources are prioritized for upload. For upload, this can potentially lead to starvation as we don't upload lower priority items until all higher priority items are uploaded.

The larger the value, higher the priority.

If set to true, logs are retained on disk via timescaledb.

If set to true, the mem buffer is written to and restored from disk.

This is useful for cases when the data needs to be persisted across logger pod restarts.

This also means that performance will be impacted as log items are written and read from disk, so only use this for event_sources that are not high throughput.

The precedence value of this instance of log options to apply.

This is different from the priority field below, which is used to determine the order in which log items are uploaded.

If there are multiple log options that match a given event source, the log options with the highest precedence value is used by the logging service. If there are multiple log options with the same precedence value, then a random one amongst the equals is used.

It is generally recommended to have higher precedence value for more specific regexes, and lower precedence value for more general regexes.

The larger the value, higher in the precedence order.

The retention time for logs retained on disk. Only used up to second precision, ignores nanoseconds.

This is only used if retain_on_disk is set to true. If this is not set, will apply the server's default retention time the first time the server encounters a matching event source.

The cluster the logs originated from.

An identifier for this LogItem that has a high probability of being unique in the context of on-prem storage. For details see 'uid' field in google.aip.dev/148. Important: The parameter should never be user defined.

Time at which this item was generated (e.g. acquisition time for sensor data). LogItems with the same event_source must be logged in increasing order of this field. Can be left unset, in which case the logging service will fill it with the time at which it wrote it (guaranteeing that it is in order).

A human readable identifier for the stream of data that this event belongs to. Should only be used for metadata that is common across multiple LogItems of this type, such as the name of the sensor that produced a reading. Information specific to this event should be added to the payload. For details see 'parent' field in google.aip.dev/148. Example: log_items/perception/frame/{camera_id}

The log item's name which is a unique identifier. The name will be created and populated when a log item is persisted in the log server. This name is what clients should use when retrieving individual items. For details see 'name' field in google.aip.dev/122. Important: The parameter should never be user defined.

workcell_name identifies the workcell/VM/dev-machine that log item belongs to. This can be used to correlate log items that don't have a corresponding deployment attribute. This field is auto populated or overridden by the logger.

===== General =====

===== Perception ===== ===== Icon =====

===== Skills =====

===== Executive =====

The cluster the blob originated from.

The organization ID prefixes the bucket that logs are uploaded into.

The event source of the blob, if any.

Used to further prefix the blob ID to prevent collisions between blobs from different event sources.

The log items to be uploaded.

The first item in the batch sets the event source and workcell name to use, and all subsequent items must have matching event source and workcell name.

Event source and workcell name cannot be empty.

The organization ID prefixes the bucket that logs are uploaded into.

The number of log items successfully ingested.

The number of log items dropped.

The unordered unique errors for the dropped log items.

Filter results by glob pattern.

Filter results by glob pattern.

Exact match only.

if unset: now - 5min

if unset: now

If set, will downsample the LogItems.

Additional key-value metadata labels to filter the returned LogItems by.

If empty, all LogItems in the query are returned. If populated, only LogItems with exact matches on all specified labels are returned.

A map of log options keys to the actual log options which are being set.

This RPC will upsert the map of log options to be applied to all current and future event sources matching the regex string in the values.

Log Options Matching: - The key of the map here is just used to identify the log options for setting/overwriting and fetching, and is not used for matching against event sources. - Instead, the LogOptions message in the value contains an event source field which should be a regex string that is used to determine which event sources the options should be applied to.

Log Option Precedence: - There might be multiple matching log options for a single event source. In that case, use the options with the highest precedence, see the log_options_precedence_value field in the LogOptions message.

Default Log Options: - Finally, if no log options are found for a given event source, then the log options for the default_event_source key if it exists are used.

• You can use this to set default log options for all event sources that don't have any specific log options set.

A description of the topics

Whether to attempt to add or remove these topics from an allowlist

The topic name/zenoh key expression

The event_sources to sync and rotate LogItems for, as regex patterns. Used only if sync_all is false.

If true, requests to sync and rotate logs of all event sources

The event_sources successfully synced and rotated LogItems for.

The event_sources that exist but were not synced and rotated due to throttling.

In bytes per second, this represents the refill rate for the token bucket.

In bytes, this represents the maximum capacity of the token bucket.

Uncompletable bag was completed with missing items (best-effort).

Default unset value

Topic is directly subscribed

Topic is subscribed through a key expression

Topic is not subscribed