Explorer orchestrator profile URL
Explorer scope
Before diving into metrics, understanding the Explorer’s scope avoids confusion. What it shows:- On-chain state as written to Arbitrum — stake, cuts, reward calls, fee earnings
- Protocol-level data: your registered service URI, your active/inactive status
- Your delegators and their individual stake amounts
- Live node health or uptime (use Prometheus for this)
- Off-chain pool worker contributions (pool workers are invisible to the protocol)
- AI model warm status or per-pipeline job counts (use tools.livepeer.cloud)
- Which gateways are currently routing work to you (use the Loki gateway API)
Active Set status
The active set is the top 100 orchestrators by total stake at each round boundary. Only active set members receive transcoding work.
- Your rank among all orchestrators
- Whether you are currently active or inactive
- Your trend over the past 30 days
Total stake
This is the sum of your self-stake and all delegated stake. It is the single most important variable for your active set ranking, and for the proportion of video transcoding work routed to you by stake-weighted gateways.Self-stake vs delegated stake
Self-stake vs delegated stake
Your self-stake is the LPT you bonded to yourself. Delegated stake comes from LPT holders who chose to delegate to you.Increase your self-stake directly by bonding more LPT. Attracting delegators requires competitive reward cut and fee share settings plus a strong performance track record — see Getting Delegates.
How stake affects job routing
How stake affects job routing
For video transcoding, stake weighting affects selection probability. Higher stake means you are more likely to be selected by stake-weighted gateways. For AI inference, routing is primarily capability-based (does your node support the requested model?) and price-based (is your price within the gateway’s limit?). Stake plays a smaller role in AI job routing.
Pending stake vs current stake
Pending stake vs current stake
Pending stake refers to delegations and undelegations that have been submitted but have not yet taken effect. Changes to stake take effect at the next round boundary.
Reward cut and fee share
These two parameters are your public service terms, visible to any potential delegator evaluating your orchestrator. What competitive looks like: Check the top 20 active orchestrators on the Explorer and compare your settings. Reward cut of 5–15% and fee share of 0–10% is typical for competitive operators. Very high reward cuts (above 30%) make delegation unattractive; delegators earn less per LPT staked than they would on a lower-cut orchestrator. Changing these parameters requires an on-chain transaction and takes effect at the next round boundary. You can update them vialivepeer_cli or directly in the Explorer interface if your wallet is connected.
Reward call history
This is the most operationally important metric to monitor regularly. Each round (~22 hours on Arbitrum), your node must callreward() to claim that round’s LPT inflation. The Explorer shows your call history round by round.
Reward calling disabled
Reward calling disabled
-reward=false disables reward calling. In split setups where the orchestrator and transcoder run as separate processes, keep the flag state consistent across every launch command. A common mistake is disabling reward on the orchestrator while a second process still owns the wallet.Check your ETH balance for gas
Check your ETH balance for gas
Reward calls require ETH on Arbitrum for gas. A dry wallet causes reward calls to fail silently. Check your orchestrator address ETH balance on Arbiscan and keep at least 0.02–0.05 ETH in the wallet as a buffer.
Check node uptime at round boundaries
Check node uptime at round boundaries
A node offline at the round boundary misses the reward call. The Arbitrum round length is approximately 22 hours. Compare regular maintenance windows against round boundaries before you dismiss the gap as random.
Using OrchestratorSiphon
Using OrchestratorSiphon
If missed rewards are a recurrent problem, consider the split setup. OrchestratorSiphon runs reward calling on a separate, stable machine independently of your GPU workload machine.
Fees earned
Cumulative ETH earned from video transcoding and AI inference fees. This is separate from LPT inflation rewards. How to read it: A low or zero fee balance despite being in the active set usually means one of three things:-
Price is too high — your
-pricePerUnitexceeds what gateways are willing to pay. Check Livepeer Explorer to compare your price to other active orchestrators. Most gateways have a maximum price they will pay, and any orchestrator above that ceiling receives no work. -
Node is unreachable — gateways cannot connect to your service URI. Test reachability from outside your network:
curl -v https://<your-service-uri>:8935/status. Port 8935 must be accessible from the public internet. -
Capabilities not registered — for AI workloads, your
aiModels.jsonconfiguration must match what gateways are requesting. Check tools.livepeer.cloud/ai/network-capabilities to confirm your pipelines appear.
Delegator list
Your delegators are LPT holders who have staked to your orchestrator. Their stake contributes to your total and increases your routing probability. What to watch:- Delegators with unbonding status — they are in the process of leaving. This reduces your stake at the next round.
- New delegations — incoming stake increasing your rank.
- Large single-delegator dependence can destabilise your active set position. One delegator holding most of your stake can move you sharply when they leave.
Service URI
Your registered service URI is what gateways use to connect to your node. It must:- Resolve to your current node’s IP or hostname
- Have port 8935 publicly accessible
- Match what your node is actually serving on
-serviceAddr <public-ip>:8935 to match what is on-chain, or update the on-chain service URI via livepeer_cli.