Data Push Data Dictionary - Dify Enterprise Docs

Field-level reference for every telemetry signal Dify Enterprise emits through Data Push. For configuring and operating the push service itself, see Data Push.

Getting Started

You do not need to read every table. Jump directly to the signal layer that matches your observability goal.

Scenario-to-Signal Map

Scenario or business need	Preferred signal	Target platform	Example signal or event
Total requests, success rate, error rate, token consumption	`Metrics` (Counter)	Prometheus / Grafana	`dify.requests.total`, `dify.tokens.total`
System throughput, latency distribution (P50/P95/P99)	`Metrics` (Histogram)	Prometheus / Grafana	`dify.workflow.duration`, `dify.message.time_to_first_token`
Why a workflow slowed down, bottleneck node, error location	`Traces` (Spans)	Jaeger / Tempo	`dify.workflow.run`, `dify.node.execution`
Detailed inputs/outputs, tool parameters, retrieval results for a specific run	`Structured Logs`	Loki / Elasticsearch	`span_detail`, `dify.tool.execution`

Telemetry Data Flow and Routing

All telemetry signals exported by Dify comply with the OpenTelemetry specification and can be split and routed by a standard OTel Collector:

Dify Enterprise Telemetry Engine (OTel Exporter)
 ├── Metrics  ──► Prometheus Exporter ──► Prometheus / Grafana Dashboard
 ├── Traces   ──► OTel / OTLP gRPC   ──► Jaeger / Grafana Tempo
 └── Logs     ──► Pod stdout (JSON) ──► your log collector ──► Loki / Elasticsearch

Transmission Rules and Resource Attributes

Resource Attributes

Global environment metadata attached to every exported OTel signal:

Attribute	Type	Example	Description
`service.name`	string	`dify`	Name of the service that produced the telemetry data
`host.name`	string	`dify-api-7f8b`	Host or Pod identifier of the producing instance

Common Identity and Correlation Attributes

Spans carry the dify.-prefixed form; metric labels and the correlation keys on logs (trace_id, span_id, tenant_id, user_id) use the bare form. Rows listing two names cover both cases.

Attribute	Type	Description
`tenant_id` / `dify.tenant_id`	string	Tenant identifier
`app_id` / `dify.app_id`	string	Application identifier; the primary dimension for filtering metrics and logs
`dify.workflow.id`	string	Workflow definition ID
`dify.workflow.run_id`	string	Unique ID for a single workflow run instance
`dify.message.id`	string	Unique message ID
`dify.conversation.id`	string	Conversation context ID (optional)
`trace_id` / `dify.trace_id`	string	OTel correlation Trace ID (32-character hex string)
`span_id`	string	OTel Span ID (16-character hex string)
`user_id` / `gen_ai.user.id`	string	End-user identifier that triggered the request (optional)
`apprunner.environment_id`	string	Deployed environments: ID of the environment for this run
`apprunner.deployment_id`	string	Deployed environments: ID of the application deployment for this run
`apprunner.revision_id`	string	Deployed environments: ID of the deployment revision used for this run

The three apprunner.* attributes apply to workflow Spans (dify.workflow.run), node Spans (dify.node.execution), and their companion logs in deployed environments. They aren’t used as metric labels.

OTel-to-Prometheus Naming Conventions

OTel signal names are automatically converted to underscore format when ingested by Prometheus. Counters gain a _total suffix; Histograms produce _bucket, _sum, and _count series, plus a unit suffix (_seconds for metrics measured in s), as the example below shows. The table below shows common examples; all other metrics follow the same rules.

OTel name	Prometheus query name
`dify.requests.total`	`dify_requests_total`
`dify.errors.total`	`dify_errors_total`
`dify.tokens.total`	`dify_tokens_total`
`dify.workflow.duration`	`dify_workflow_duration_seconds_bucket` and related Histogram series

Null Values and Content-Gated Fields

Null value behavior: attributes with no value are omitted from Trace Spans; in structured log JSON output they are explicitly rendered as null.
Content-gating: attributes that may hold sensitive text (Input/Output/Query/Prompt) are gated by ENTERPRISE_INCLUDE_CONTENT. When content push is disabled (the default), the value is replaced by a reference string (e.g. ref:{id_type}={uuid}) rather than null.
High-cardinality warning: tenant_id, app_id, and trace_id are high-cardinality dimensions. Do not store unbounded text fields as Prometheus labels.

Traces and Spans

Traces present the complete time sequence and nesting topology of a single call. The OTel trace_id is derived deterministically from the business run ID, so every signal of one run shares it; query Jaeger or Tempo by this trace_id. The dify.trace_id attribute keeps the original business UUID (for example the workflow run ID) for joining with business data.

dify.workflow.run

Attribute	Type	Description
`dify.trace_id`	string	Business Trace ID (workflow run ID)
`dify.tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.workflow.id`	string	Workflow definition ID
`dify.workflow.run_id`	string	Unique ID for this run
`dify.workflow.status`	string	`succeeded`, `failed`, `stopped`, etc.
`dify.workflow.error`	string	Error message on failure
`dify.workflow.elapsed_time`	float	Total execution time (seconds)
`dify.invoke_from`	string	`api`, `webapp`, `debug`
`dify.conversation.id`	string	Conversation ID (optional)
`dify.message.id`	string	Message ID (optional)
`dify.invoked_by`	string	User ID that triggered the run
`gen_ai.usage.total_tokens`	int	Total token count across all nodes (optional)
`gen_ai.user.id`	string	End-user identifier (optional)
`dify.parent.trace_id`	string	Parent workflow trace ID (optional)
`dify.parent.workflow.run_id`	string	Parent workflow run ID (optional)
`dify.parent.node.execution_id`	string	Parent node execution ID (optional)
`dify.parent.app.id`	string	Parent application ID (optional)

dify.node.execution

Attribute	Type	Description
`dify.trace_id`	string	Business Trace ID
`dify.tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.workflow.id`	string	Workflow definition ID
`dify.workflow.run_id`	string	Workflow run ID
`dify.message.id`	string	Message ID (optional)
`dify.conversation.id`	string	Conversation ID (optional)
`dify.node.execution_id`	string	Unique node execution ID
`dify.node.id`	string	Node ID in the Workflow graph
`dify.node.type`	string	Node type (see Appendix)
`dify.node.title`	string	Display title
`dify.node.status`	string	`succeeded`, `failed`
`dify.node.error`	string	Error message on failure
`dify.node.elapsed_time`	float	Execution time (seconds)
`dify.node.index`	int	Execution sequence index
`dify.node.predecessor_node_id`	string	ID of the node that triggered this node
`dify.node.iteration_id`	string	Iteration ID (optional)
`dify.node.loop_id`	string	Loop ID (optional)
`dify.node.parallel_id`	string	Parallel branch ID (optional)
`dify.node.invoked_by`	string	User ID that triggered execution
`gen_ai.usage.input_tokens`	int	Prompt token count (LLM nodes only)
`gen_ai.usage.output_tokens`	int	Completion token count (LLM nodes only)
`gen_ai.usage.total_tokens`	int	Total token count (LLM nodes only)
`gen_ai.request.model`	string	LLM model name (LLM nodes only)
`gen_ai.provider.name`	string	LLM provider name (LLM nodes only)
`gen_ai.user.id`	string	End-user identifier (optional)

dify.node.execution.draft

Carries the same attributes as dify.node.execution, but is produced only during a single-node test or a draft preview run in the Studio canvas. Use this signal to separate production telemetry from test runs.

Metrics

Counters

All counters are cumulative and are never sampled: the trace sampling rate (ENTERPRISE_OTEL_SAMPLING_RATE, see the environment variable reference) does not affect them.

Token Counters

Metric	Unit	Description
`dify.tokens.total`	`{token}`	Total tokens consumed
`dify.tokens.input`	`{token}`	Input (prompt) tokens
`dify.tokens.output`	`{token}`	Output (completion) tokens

Labels: tenant_id, app_id, operation_type, model_provider, model_name, node_type (for node_execution only)

The workflow-level dify.tokens.total already includes tokens from all node_execution operations. Filter by operation_type to avoid double-counting.

Token hierarchy and query patterns Token metrics are emitted at multiple levels. Understanding the hierarchy prevents double-counting:

App-level total
├── workflow          ← sum of all node_execution tokens (do not add together)
│   └── node_execution ← per-node breakdown
├── message           ← standalone (non-workflow chat apps only)
├── rule_generate     ← standalone auxiliary LLM call
├── code_generate     ← standalone auxiliary LLM call
├── structured_output ← standalone auxiliary LLM call
└── instruction_modify← standalone auxiliary LLM call

Key rule: workflow tokens already include all node_execution tokens. Never add both together.

Available labels on token metrics: tenant_id, app_id, operation_type, model_provider, model_name, node_type.
Application name is carried on the trace side only, as dify.app.name on the span companion logs (see Span Companion Logs), never as a metric label. Use app_id in metric queries.
Cost exists only on the node companion logs (dify.node.total_price, see Span Companion Logs); there is no cost metric.

Request Counter

Metric	Unit	Description
`dify.requests.total`	`{request}`	Total operation count

The type label identifies the operation (for example dify_requests_total{type="workflow"}); additional labels vary by type:

Type	Additional labels
`workflow`	`tenant_id`, `app_id`, `status`, `invoke_from`
`node`	`tenant_id`, `app_id`, `node_type`, `model_provider`, `model_name`, `status`
`draft_node`	`tenant_id`, `app_id`, `node_type`, `model_provider`, `model_name`, `status`
`message`	`tenant_id`, `app_id`, `model_provider`, `model_name`, `status`, `invoke_from`
`tool`	`tenant_id`, `app_id`, `tool_name`
`moderation`	`tenant_id`, `app_id`
`suggested_question`	`tenant_id`, `app_id`, `model_provider`, `model_name`
`dataset_retrieval`	`tenant_id`, `app_id`
`generate_name`	`tenant_id`, `app_id`
`prompt_generation`	`tenant_id`, `app_id`, `operation_type`, `model_provider`, `model_name`, `status`

Error Counter

Metric	Unit	Description
`dify.errors.total`	`{error}`	Total failed operation count

The error counter uses the same type label; additional labels vary by type:

Type	Additional labels
`workflow`	`tenant_id`, `app_id`
`node`	`tenant_id`, `app_id`, `node_type`, `model_provider`, `model_name`
`draft_node`	`tenant_id`, `app_id`, `node_type`, `model_provider`, `model_name`
`message`	`tenant_id`, `app_id`, `model_provider`, `model_name`
`tool`	`tenant_id`, `app_id`, `tool_name`
`prompt_generation`	`tenant_id`, `app_id`, `operation_type`, `model_provider`, `model_name`

Other Counters

Metric	Unit	Labels
`dify.feedback.total`	`{feedback}`	`tenant_id`, `app_id`, `rating`
`dify.dataset.retrievals.total`	`{retrieval}`	`tenant_id`, `app_id`, `dataset_id`, `embedding_model_provider`, `embedding_model`, `rerank_model_provider`, `rerank_model`
`dify.app.created.total`	`{app}`	`tenant_id`, `app_id`, `mode`
`dify.app.updated.total`	`{app}`	`tenant_id`, `app_id`
`dify.app.deleted.total`	`{app}`	`tenant_id`, `app_id`

Histograms

Unlike Counters, which can only produce averages (which extreme values can distort), Histograms support calculations for the P50, P95, and P99 percentiles that more accurately describe the actual distribution.Each Histogram in Prometheus expands into three series: _bucket, _sum, and _count. P95/P99 may return NaN in low-traffic time windows; increase the query window if this occurs. When latency is concentrated within 5 seconds, such as for TTFT or tool calls, percentile precision is limited. For higher-precision evaluation, extend the query window or contact the Dify team for bucket configuration details.

Metric	Unit	Labels
`dify.workflow.duration`	`s`	`tenant_id`, `app_id`, `status`
`dify.node.duration`	`s`	`tenant_id`, `app_id`, `node_type`, `model_provider`, `model_name`, `plugin_name`
`dify.message.duration`	`s`	`tenant_id`, `app_id`, `model_provider`, `model_name`
`dify.message.time_to_first_token`	`s`	`tenant_id`, `app_id`, `model_provider`, `model_name`
`dify.tool.duration`	`s`	`tenant_id`, `app_id`, `tool_name`
`dify.prompt_generation.duration`	`s`	`tenant_id`, `app_id`, `operation_type`, `model_provider`, `model_name`

Structured Logs

Structured logs provide detailed business data. Logs fall into two categories: Span companion logs and standalone business logs.

ENTERPRISE_OTEL_SAMPLING_RATE only affects the number of Spans sent to Trace backends such as Jaeger and Tempo. Structured logs (span_detail and metric_only) are always written to Pod standard output at 100% and are not controlled by the sampling rate.

Span Companion Logs

Signal type: span_detail. These logs are tightly coupled to a Trace Span (sharing trace_id and span_id) and contain content-gated detailed context.

dify.workflow.run companion log

Common attributes: all Span attributes (see Traces section) plus: Event attributes:

dify.event.name: dify.workflow.run
dify.event.signal: span_detail
trace_id, span_id, tenant_id, user_id

Additional attribute	Type	Always present	Description
`dify.app.name`	string	No	Application display name
`dify.workspace.name`	string	No	Workspace display name
`dify.workflow.version`	string	Yes	Workflow definition version
`dify.workflow.inputs`	string/JSON	Yes	Input parameters (content-gated)
`dify.workflow.outputs`	string/JSON	Yes	Output results (content-gated)
`dify.workflow.query`	string	No	User query text (content-gated)

dify.node.execution and dify.node.execution.draft companion logs

Common attributes: all Span attributes (see Traces section) plus: Event attributes:

dify.event.name: dify.node.execution or dify.node.execution.draft
dify.event.signal: span_detail
trace_id, span_id, tenant_id, user_id

Additional attribute	Type	Always present	Description
`dify.app.name`	string	No	Application display name
`dify.workspace.name`	string	No	Workspace display name
`dify.invoke_from`	string	No	Invocation source
`gen_ai.tool.name`	string	No	Tool name (tool nodes only)
`dify.node.total_price`	float	No	Cost (LLM nodes only)
`dify.node.currency`	string	No	Currency code (LLM nodes only)
`dify.node.iteration_index`	int	No	Iteration index (iteration nodes)
`dify.node.loop_index`	int	No	Loop index (loop nodes)
`dify.plugin.name`	string	No	Plugin name (tool / knowledge nodes)
`dify.credential.name`	string	No	Credential name (plugin nodes)
`dify.credential.id`	string	No	Credential ID (plugin nodes)
`dify.dataset.ids`	JSON array	No	Dataset IDs (knowledge nodes)
`dify.dataset.names`	JSON array	No	Dataset names (knowledge nodes)
`dify.node.inputs`	string/JSON	Yes	Node inputs (content-gated)
`dify.node.outputs`	string/JSON	Yes	Node outputs (content-gated)
`dify.node.process_data`	string/JSON	No	Processing data (content-gated)

Standalone Logs

Logs without a structured Span. Their signal type is metric_only: the event feeds counters and carries its detail in this log, with no companion span.

dify.message.run

Attribute	Type	Description
`dify.event.name`	string	`dify.message.run`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID (32-character hex)
`span_id`	string	OTel Span ID (16-character hex)
`tenant_id`	string	Tenant identifier
`user_id`	string	User identifier (optional)
`dify.app_id`	string	Application identifier
`dify.message.id`	string	Message identifier
`dify.conversation.id`	string	Conversation ID (optional)
`dify.workflow.run_id`	string	Workflow run ID (optional)
`dify.invoke_from`	string	`service-api`, `web-app`, `debugger`, `explore`
`gen_ai.provider.name`	string	LLM provider
`gen_ai.request.model`	string	LLM model
`gen_ai.usage.input_tokens`	int	Input token count
`gen_ai.usage.output_tokens`	int	Output token count
`gen_ai.usage.total_tokens`	int	Total token count
`dify.message.status`	string	`succeeded`, `failed`
`dify.message.error`	string	Error message (if failed)
`dify.message.duration`	float	Duration (seconds)
`dify.message.time_to_first_token`	float	Time to first token / TTFT (seconds)
`dify.message.inputs`	string/JSON	Inputs (content-gated)
`dify.message.outputs`	string/JSON	Outputs (content-gated)

dify.tool.execution

Attribute	Type	Description
`dify.event.name`	string	`dify.tool.execution`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID
`span_id`	string	OTel Span ID
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.message.id`	string	Message identifier
`dify.tool.name`	string	Tool name
`dify.tool.duration`	float	Duration (seconds)
`dify.tool.status`	string	`succeeded`, `failed`
`dify.tool.error`	string	Error message (if failed)
`dify.tool.inputs`	string/JSON	Inputs (content-gated)
`dify.tool.outputs`	string/JSON	Outputs (content-gated)
`dify.tool.parameters`	string/JSON	Parameters (content-gated)
`dify.tool.config`	string/JSON	Configuration (content-gated)

dify.moderation.check

Attribute	Type	Description
`dify.event.name`	string	`dify.moderation.check`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID
`span_id`	string	OTel Span ID
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.message.id`	string	Message identifier
`dify.moderation.type`	string	`input`, `output`
`dify.moderation.action`	string	`pass`, `block`, `flag`
`dify.moderation.flagged`	boolean	Whether the content was flagged
`dify.moderation.categories`	JSON array	Categories that triggered the flag
`dify.moderation.query`	string	Content (content-gated)

dify.suggested_question.generation

Attribute	Type	Description
`dify.event.name`	string	`dify.suggested_question.generation`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID
`span_id`	string	OTel Span ID
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.message.id`	string	Message identifier
`dify.suggested_question.count`	int	Number of questions generated
`dify.suggested_question.duration`	float	Duration (seconds)
`dify.suggested_question.status`	string	`succeeded`, `failed`
`dify.suggested_question.error`	string	Error message (if failed)
`dify.suggested_question.questions`	JSON array	Generated questions (content-gated)

dify.dataset.retrieval

Attribute	Type	Description
`dify.event.name`	string	`dify.dataset.retrieval`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID
`span_id`	string	OTel Span ID
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.message.id`	string	Message identifier
`dify.dataset.id`	string	Dataset identifier
`dify.dataset.name`	string	Dataset name
`dify.dataset.embedding_providers`	JSON array	Embedding model provider (one per dataset)
`dify.dataset.embedding_models`	JSON array	Embedding model (one per dataset)
`dify.retrieval.rerank_provider`	string	Rerank model provider
`dify.retrieval.rerank_model`	string	Rerank model name
`dify.retrieval.query`	string	Search query (content-gated)
`dify.retrieval.document_count`	int	Number of documents retrieved
`dify.retrieval.duration`	float	Duration (seconds)
`dify.retrieval.status`	string	`succeeded`, `failed`
`dify.retrieval.error`	string	Error message (if failed)
`dify.dataset.documents`	JSON array	Documents (content-gated)

dify.generate_name.execution

Attribute	Type	Description
`dify.event.name`	string	`dify.generate_name.execution`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID
`span_id`	string	OTel Span ID
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.conversation.id`	string	Conversation identifier
`dify.generate_name.duration`	float	Duration (seconds)
`dify.generate_name.status`	string	`succeeded`, `failed`
`dify.generate_name.error`	string	Error message (if failed)
`dify.generate_name.inputs`	string/JSON	Inputs (content-gated)
`dify.generate_name.outputs`	string	Generated name (content-gated)

dify.prompt_generation.execution

Attribute	Type	Description
`dify.event.name`	string	`dify.prompt_generation.execution`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID
`span_id`	string	OTel Span ID
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.prompt_generation.operation_type`	string	Operation type (see Appendix)
`gen_ai.provider.name`	string	LLM provider
`gen_ai.request.model`	string	LLM model
`gen_ai.usage.input_tokens`	int	Input token count
`gen_ai.usage.output_tokens`	int	Output token count
`gen_ai.usage.total_tokens`	int	Total token count
`dify.prompt_generation.duration`	float	Duration (seconds)
`dify.prompt_generation.status`	string	`succeeded`, `failed`
`dify.prompt_generation.error`	string	Error message (if failed)
`dify.prompt_generation.instruction`	string	Instruction (content-gated)
`dify.prompt_generation.output`	string/JSON	Output (content-gated)

dify.app.created

Attribute	Type	Description
`dify.event.name`	string	`dify.app.created`
`dify.event.signal`	string	`metric_only`
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.app.mode`	string	`chat`, `completion`, `agent-chat`, `workflow`
`dify.app.created_at`	string	Timestamp (ISO 8601)

dify.app.updated

Attribute	Type	Description
`dify.event.name`	string	`dify.app.updated`
`dify.event.signal`	string	`metric_only`
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.app.updated_at`	string	Timestamp (ISO 8601)

dify.app.deleted

Attribute	Type	Description
`dify.event.name`	string	`dify.app.deleted`
`dify.event.signal`	string	`metric_only`
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.app.deleted_at`	string	Timestamp (ISO 8601)

dify.feedback.created

Attribute	Type	Description
`dify.event.name`	string	`dify.feedback.created`
`dify.event.signal`	string	`metric_only`
`trace_id`	string	OTel Trace ID
`span_id`	string	OTel Span ID
`tenant_id`	string	Tenant identifier
`dify.app_id`	string	Application identifier
`dify.message.id`	string	Message identifier
`dify.feedback.rating`	string	`like`, `dislike`, `null`
`dify.feedback.content`	string	Feedback text (content-gated)
`dify.feedback.created_at`	string	Timestamp (ISO 8601)

dify.telemetry.rehydration_failed

A diagnostic event for telemetry pipeline health, emitted when a buffered telemetry payload cannot be reconstructed (rehydrated) for processing. A sustained rate indicates telemetry data loss.

Attribute	Type	Description
`dify.event.name`	string	`dify.telemetry.rehydration_failed`
`dify.event.signal`	string	`metric_only`
`tenant_id`	string	Tenant identifier
`dify.telemetry.error`	string	Error message
`dify.telemetry.payload_type`	string	Payload type (see Appendix)
`dify.telemetry.correlation_id`	string	Correlation ID

Content-Gated Attributes

The following fields may contain user inputs, model outputs, document content, tool parameters, or feedback text. Raw values are transmitted only when content push is enabled on the deployment: ENTERPRISE_INCLUDE_CONTENT (default false; see the environment variable reference). When disabled, these attributes are replaced with ref:{id_type}={uuid}. The id_type names the correlating entity (for example workflow_run_id, node_execution_id, or message_id); look up that record in the Dify database to retrieve the original content.

Attribute	Signal
`dify.workflow.inputs`	`dify.workflow.run`
`dify.workflow.outputs`	`dify.workflow.run`
`dify.workflow.query`	`dify.workflow.run`
`dify.node.inputs`	`dify.node.execution`
`dify.node.outputs`	`dify.node.execution`
`dify.node.process_data`	`dify.node.execution`
`dify.message.inputs`	`dify.message.run`
`dify.message.outputs`	`dify.message.run`
`dify.tool.inputs`	`dify.tool.execution`
`dify.tool.outputs`	`dify.tool.execution`
`dify.tool.parameters`	`dify.tool.execution`
`dify.tool.config`	`dify.tool.execution`
`dify.moderation.query`	`dify.moderation.check`
`dify.suggested_question.questions`	`dify.suggested_question.generation`
`dify.retrieval.query`	`dify.dataset.retrieval`
`dify.dataset.documents`	`dify.dataset.retrieval`
`dify.generate_name.inputs`	`dify.generate_name.execution`
`dify.generate_name.outputs`	`dify.generate_name.execution`
`dify.prompt_generation.instruction`	`dify.prompt_generation.execution`
`dify.prompt_generation.output`	`dify.prompt_generation.execution`
`dify.feedback.content`	`dify.feedback.created`

dify.node.execution.draft carries the same gated fields as dify.node.execution.

In production, it is recommended to disable content push by default. Enable it in a controlled environment only after confirming data classification, access controls, redaction requirements, retention periods, and audit obligations.

Appendix

Operation Types

workflow, node_execution, message, rule_generate, code_generate, structured_output, instruction_modify These values belong to the operation_type label on token metrics and to dify.prompt_generation.operation_type. The request and error counters use the separate type label; its values are listed under Counters.

Node Types

start, end, answer, llm, knowledge-retrieval, knowledge-index, if-else, code, template-transform, question-classifier, http-request, tool, datasource, variable-aggregator, loop, iteration, parameter-extractor, assigner, document-extractor, list-operator, agent, trigger-webhook, trigger-schedule, trigger-plugin, human-input

Workflow Statuses

running, succeeded, failed, stopped, partial-succeeded, paused

Payload Types

workflow, node, message, tool, moderation, suggested_question, dataset_retrieval, generate_name, prompt_generation, app, feedback

Invoke From Values

Workflow spans (dify.invoke_from on dify.workflow.run): api, webapp, debug
Message logs (dify.invoke_from on dify.message.run): service-api, web-app, debugger, explore

Null Value Behavior

Spans: attributes with null values are omitted.
Logs: attributes with null values are rendered as null in the JSON output.
Content-gated fields: replaced by a reference string rather than set to null.

​Getting Started

​Scenario-to-Signal Map

​Telemetry Data Flow and Routing

​Transmission Rules and Resource Attributes

​Resource Attributes

​Common Identity and Correlation Attributes

​OTel-to-Prometheus Naming Conventions

​Null Values and Content-Gated Fields

​Traces and Spans

​dify.workflow.run

​dify.node.execution

​dify.node.execution.draft

​Metrics

​Counters

​Token Counters

​Request Counter

​Error Counter

​Other Counters

​Histograms

​Structured Logs

​Span Companion Logs

​dify.workflow.run companion log

​dify.node.execution and dify.node.execution.draft companion logs

​Standalone Logs

​dify.message.run

​dify.tool.execution

​dify.moderation.check

​dify.suggested_question.generation

​dify.dataset.retrieval

​dify.generate_name.execution

​dify.prompt_generation.execution

​dify.app.created

​dify.app.updated

​dify.app.deleted

​dify.feedback.created

​dify.telemetry.rehydration_failed

​Content-Gated Attributes

​Appendix

​Operation Types

​Node Types

​Workflow Statuses

​Payload Types

​Invoke From Values

​Null Value Behavior

Getting Started

Scenario-to-Signal Map

Telemetry Data Flow and Routing

Transmission Rules and Resource Attributes

Resource Attributes

Common Identity and Correlation Attributes

OTel-to-Prometheus Naming Conventions

Null Values and Content-Gated Fields

Traces and Spans

dify.workflow.run

dify.node.execution

dify.node.execution.draft

Metrics

Counters

Token Counters

Request Counter

Error Counter

Other Counters

Histograms

Structured Logs

Span Companion Logs

dify.workflow.run companion log

dify.node.execution and dify.node.execution.draft companion logs

Standalone Logs

dify.message.run

dify.tool.execution

dify.moderation.check

dify.suggested_question.generation

dify.dataset.retrieval

dify.generate_name.execution

dify.prompt_generation.execution

dify.app.created

dify.app.updated

dify.app.deleted

dify.feedback.created

dify.telemetry.rehydration_failed

Content-Gated Attributes

Appendix

Operation Types

Node Types

Workflow Statuses

Payload Types

Invoke From Values

Null Value Behavior