IN the agents dashboard I see that one agent has 2 workers assigned, yet when I click on it I always see one. which one is true?

Thanks for reaching out, luisa. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

I only see the ready ones there

Thanks — that helps. You're correct: the Workers table only shows workers in the READY/active state. The tile’s “WORKERS 2” can include workers in other lifecycle states (starting up, draining, or recently disconnected) or reflect a slightly different refresh interval. What to do next: • Wait ~1–2 minutes and refresh; counts usually sync once transitional workers finish starting/draining. • Check for recent deployments or restarts (those create transient workers). • Watch the STATUS column for non-READY states (e.g., STARTING, DRAINING) — those explain the extra count. If the mismatch persists for more than a few minutes and there were no recent deployments, tell me how long it’s been and whether you want me to walk through further checks. , with interactive elements

is there a better way to see the all of them in real time?

Thanks — good question. You're asking for a way to see all workers (including non-READY/transitional ones) in real time. What you can do now: • Use the Agent Console (Launch Console from the Agents dashboard) — it shows live state and events for workers and sessions in real time. • Use the CLI to list agent instances and status:

lk agent list

and

lk agent status

(shows replica counts and lifecycle states). • Watch the Workers panel for STATUS changes (STARTING, DRAINING, READY); give the UI ~1–2 minutes to refresh if you just deployed. Quick clarification: do you have access to the Agent Console and the

lk

CLI? If so I’ll give the exact steps for your environment. Sources: examples/drive-thru | livekit/livekit_composite | DeepWiki | Agent deployment overview | LiveKit Documentation | Agent Console | LiveKit Documentation, with interactive elements

I don’t how can I install the lk cli

Thanks — I’ll walk you through installing the

lk

CLI and the minimal steps to use it to view all workers in real time. Pick your OS and follow the steps: 1) macOS (Homebrew) • Install: brew install livekit-cli • Verify: lk --help 2) Linux (script) • Install: curl -sSL https://get.livekit.io/cli | bash • Verify: lk --help 3) Windows (winget) • Install: winget install LiveKit.LiveKitCLI • Verify: lk --help 4) Alternative (any OS) • Download a prebuilt binary from the GitHub releases page: https://github.com/livekit/livekit-cli/releases/latest • Make it executable and put it on your PATH, then run

lk --help

. After install — link the CLI to your LiveKit Cloud project: • Authenticate: lk cloud auth (this opens a browser to sign in and link your project) Useful commands to see all workers / real-time state: • List agents: lk agent list • Show agent status (includes replica counts and lifecycle states): lk agent status • Stream logs for debugging: lk agent logs --follow If you tell me your OS, I’ll give the exact command you should run and help with any errors you hit. Sources: Google AI and LiveKit | LiveKit Documentation | examples/drive-thru | livekit/livekit_composite | DeepWiki | Agents (Python) | livekit/livekit_composite | DeepWiki, with interactive elements

I don’t see any agent if tho I connected to that correct project and in the dashboard I see them we are using self hosted agents

Thanks — that helps clarify things. Short answer: the

lk

CLI only shows Cloud-hosted agents. For self-hosted agents, the CLI won’t list them even when you’re connected to the correct project. The dashboard’s observability features are what show your self-hosted agents, and direct access to the hosts is required for more realtime/detail. What you can do next: • Use the LiveKit dashboard’s Agent Console and Workers panel for observability (this will show your self-hosted agents). • Tail the agent logs on the machines/cluster where your agents run (set

--log-level=DEBUG

or use your host’s log streaming). • Ensure agent observability is enabled in your project settings so the dashboard receives telemetry. • If you want a programmatic view, add log/metrics forwarding from your hosts (e.g., stdout → ELK/CloudWatch/Prometheus) so you can see all lifecycle states in real time. Would you like exact commands for streaming logs from your agent host (tell me your host type: Docker, Kubernetes, or VM)? Sources: Where to find your Agent Logs | LiveKit | Server-side | livekit/livekit_composite | DeepWiki | Why is my Agent insights data missing? | LiveKit | Self-hosted deployments | LiveKit Documentation, with interactive elements

oh that’s sad

I hear you — that is disappointing. I can help make monitoring as close to real‑time as possible for your self‑hosted agents. Options I can help set up right now: • Tail agent logs on your hosts (Docker, Kubernetes, or VM)—I can give exact commands for your environment. • Forward logs/metrics to a central system (CloudWatch/ELK/Prometheus) and show what metrics to capture. • Configure higher log verbosity (

--log-level=DEBUG

) and where to expose health/state endpoints so the dashboard picks up transitions faster. • Walk through enabling Agent Observability in the dashboard if it’s not already enabled. Which host type are your agents running on (Docker, Kubernetes, or VM)? I’ll give the exact commands and a short recommended setup. , with interactive elements

how can I see the agent dispatch errors? from this graph

Thanks — you’re asking how to view and investigate the “Agent dispatch errors” shown in that red segment of the graph. What those errors mean • Agent dispatch errors occur when the system fails to assign an agent to a session/room (examples: no available workers, timeouts connecting to a worker, or worker crashes). Where to get detailed info • Agent logs on the hosts: the most direct place to see dispatch error details and stack traces. For self‑hosted agents you must tail the logs on the machines/cluster running the agents. • Sessions → Agent insights → Logs in the dashboard: shows session-level observability when enabled (note: logs that never result in a session may not appear here). • External logging/metrics: forward agent logs to CloudWatch/ELK/Datadog/Sentry to capture dispatch failures across hosts and query them centrally. Actions to investigate right now 1. Check agent logs around the timestamps that show red errors (use your host’s log streaming: Docker/K8s/VM). 2. Confirm worker availability at those times (were workers STARTING or DRAINING?). 3. Look for connection timeout messages like “room connection was not established within 10 seconds after calling job_entry” or worker crash/oom errors. 4. If you use dashboards/alerts, query logs by the graph’s interval (e.g., 16 Apr, 0900–1700) to correlate sessions vs dispatch failures. If you tell me how your agents are hosted (Docker, Kubernetes, or VM) and which timestamps you want to inspect, I’ll give exact log commands and the key log lines to look for. Sources: Agent dispatch | LiveKit Documentation | How to Get Help from LiveKit | Agents Framework | livekit/livekit_composite | DeepWiki | Where to find your Agent Logs | LiveKit, with interactive elements