Diagnostics

The Diagnostics tab gives you visibility into the running system through log viewing, on-demand state snapshots, and agent fleet snapshots. Use it first when troubleshooting ingest failures, performance issues, or unexpected agent behavior.

Logs are fetched on demand when you select a log type. They are not streamed in real time.

Log Viewer

The Log Viewer displays log content in a scrollable, read-only text area. For very large log files, only the most recent 200 lines are returned.

Log type	Content
Backend	The console application log. This is the primary source for ingest errors, database communication failures, job engine events, and exception stack traces. The default log level is INFO.
Frontend	Browser console output captured by the console UI. Use it to diagnose UI errors that do not produce a visible message. The log can be empty if no console errors have occurred since the page was last loaded.
Application Log	Periodic system snapshots, one JSON entry per snapshot, written to data/logs/application.log.
Agent Dashboard Log	Periodic agent fleet snapshots, one JSON entry per snapshot, written to data/logs/agentdashboard.jsonl.

System Snapshot Log

The System Snapshot Log records the state of the whole system at a configurable interval. Each entry is a self-contained JSON object with a timestamp and the following data:

• JVM heap, CPU, thread, and garbage-collection metrics
• API call counts and average latency since the last restart
• Database connectivity and round-trip latency
• Database size and row counts
• Ingest lag per cluster (time between the latest ingested event and now)
• Disk read and write latency
• Analytics database version, query rate, and storage statistics
• Configuration database health: ping latency, active connections, pool utilization, and size

Controls

• Trigger Snapshot — writes one entry immediately, regardless of the configured interval. Run it before reporting a support issue to capture a consistent point-in-time picture.
• Snapshot interval — how often a snapshot is written automatically. Range: 10 to 3600 seconds. Default: 30 seconds. A change takes effect without a restart.

Agent Dashboard Snapshot Log

The Agent Dashboard Snapshot Log records the state of every registered agent at a configurable interval. The registered fleet consists of the data-classification pipeline agents (extraction, classification, and linguistic coherence, or LC) and the local console entry.

Each entry contains a fleet summary and one record per agent. All per-agent values come from the agent's most recent heartbeat. An offline agent shows its last known values.

Fleet summary fields

Field	Content
ts / iso	Timestamp as epoch milliseconds and ISO-8601
agentCount	Total registered agents
onlineCount	Agents with status ONLINE
offlineCount	Agents with status OFFLINE or DEGRADED

Per-agent fields

Field	Content
agentId / name	Agent ID and display name
type	PIPELINE_EXTRACTION, PIPELINE_CLASSIFY, PIPELINE_LC, or TRINO
status	ONLINE, OFFLINE, or DEGRADED
version	Agent software version
hostname / ip	Host reported by the agent
lastSeenMs	Time of the last heartbeat, in epoch milliseconds
chConnected	Whether the agent can reach the analytics database directly
draining	Whether the agent is in maintenance drain mode (no new work is assigned)
siteId	Optional site or location label
upgradeStatus	IDLE, PENDING, UPGRADING, DONE, or FAILED
hostCpuPct / jvmCpuPct	Host CPU and JVM process CPU utilization, in percent
heapUsedPct	JVM heap used, in percent
heapUsedMb / heapMaxMb	JVM heap used and maximum, in MB
threadCount	JVM thread count

Pipeline throughput fields

Pipeline agents report these additional fields.

Field	Content
pipelineWorkInHand	Work items the agent has claimed but not completed
pipelineIngressInflight / pipelineIngressCapacity	Current in-flight ingress requests and the agent's ingress capacity
pipeline503Window60s	Ingress back-pressure (503) responses in the last 60 seconds
pipelineFilesProcessedLifetime	Files processed by the agent since it started
analyzerUp	Whether the agent's analyzer subprocess (LC or classification) is running
analyzerRestarts	Analyzer subprocess restarts since the agent started
analyzerUptimeSec	Uptime of the current analyzer subprocess, in seconds
analyzerLastDeathReason	Reason for the most recent analyzer subprocess exit (empty if none)

Controls

• Trigger Snapshot — writes one entry immediately, regardless of the configured interval.
• Snapshot interval — how often a snapshot is written automatically. Range: 10 to 3600 seconds. Default: 60 seconds.

Use cases

• Capacity trending — track pipelineWorkInHand and pipelineIngressInflight over time as scan load increases.
• Back-pressure detection — a rising pipeline503Window60s means an agent is at capacity and shedding requests. Scale out that pipeline stage.
• Connectivity issues — chConnected shows which agents lost their path to the analytics database, and when.
• Analyzer health — analyzerUp, analyzerRestarts, and analyzerLastDeathReason reveal crashes or restart loops.
• Upgrade auditing — upgradeStatus and version track the rollout of new agent builds.
• Post-incident analysis — compare pipelineWorkInHand, pipeline503Window60s, and analyzerRestarts around the incident window to see when a stage saturated or an analyzer restarted.

Workflows

View recent backend errors

1. Select Backend in the log type selector.
2. Scroll to the bottom. The most recent entries appear last.
3. Search the page (Ctrl+F or Cmd+F) for ERROR or WARN.

Look for lines that contain a job ID or device name when troubleshooting ingest or connectivity problems.

Capture a system snapshot

1. Click Trigger Snapshot in the System Snapshot Log card.
2. Wait for the confirmation message.
3. Select Application Log in the log type selector. The new entry appears at the end of the log.

Compare consecutive entries to see how job state or memory changed between two points in time.

Capture an agent fleet snapshot

1. Click Trigger Snapshot in the Agent Dashboard Snapshot Log card.
2. Wait for the confirmation message.
3. Select Agent Dashboard Log in the log type selector. The new entry shows the state of all registered agents at that moment.