Aller au contenu principal

Operating a Hydra Node

This page aims at helping Hydra users troubleshoot issues when running their own instances of hydra-node and participate in a Hydra Head.


Following ADR-9 design principles, the hydra-node provides JSON formatted logs on the stdout stream, one line per log item. The log items follow a JSON schema. This logging capability is kept voluntarily simple and non configurable in order to ease integration of Hydra logging into more general log analysis infrastructure, whether a custom ELK stack, third-party services, docker sidecars...


There is an unpublished log-filter executable that one can attach to a hydra-node in order to trim down the volume of information in the log stream. This filter provides some filtering features, namely removing transactions bodies and replacing them with transaction ids, but it's not general enough to warrant publication. Similar capabilities can be easily provided with tools like jq.


When given --monitoring-port PORT argument, the hydra-node executable will expose a Prometheus compatible HTTP /metrics endpoint on the given port to enable scraping of exposed metrics.

For example, assuming a hydra-node was started with --monitoring-port 6001, this command

curl http://localhost:6001/metrics

will output

# TYPE hydra_head_confirmed_tx counter
hydra_head_confirmed_tx 0
# TYPE hydra_head_events counter
hydra_head_events 50467
# TYPE hydra_head_requested_tx counter
hydra_head_requested_tx 0
# TYPE hydra_head_tx_confirmation_time_ms histogram
hydra_head_tx_confirmation_time_ms_bucket{le="5.0"} 0.0
hydra_head_tx_confirmation_time_ms_bucket{le="10.0"} 0.0
hydra_head_tx_confirmation_time_ms_bucket{le="50.0"} 0.0
hydra_head_tx_confirmation_time_ms_bucket{le="100.0"} 0.0
hydra_head_tx_confirmation_time_ms_bucket{le="1000.0"} 0.0
hydra_head_tx_confirmation_time_ms_bucket{le="+Inf"} 0.0
hydra_head_tx_confirmation_time_ms_sum 0.0
hydra_head_tx_confirmation_time_ms_count 0

Common Issues

No Head is observed from the chain

  • hydra-node is connected to a cardano-node that's on the wrong network. Check the --network command-line argument and the cardano-node configuration
  • Note that the hydra-node cannot start if it cannot connect to the cardano-node, which might require some time as the cardano-node needs to revalidate its database and possibly even reconstruct its ledger state when it starts and its connections are not open until it's ready. If running as a service or a container, make sure the orchestrator restarts the process when it crashes
  • The Scripts transaction identifier is invalid. This transaction id is available in the release page for the 3 major networks (preview, preprod, mainnet)
  • The hydra-node's Cardano signing key is inconsistent with the Verification key from the Init transaction. Check the --cardano-signing-key parameter points to the right key, and that peers have the correct --cardano-verification-key for your host.
  • The peers' Cardano verification keys are incorrect. This is mirroring the above issue, check parameters on all peers.

Head does not make progress

  • Peers are not correctly connected to each others'. Check the --peer arguments point to the right host:port for each peer. PeerConnected message should be sent to the client (or appears in the logs) and be consistent for all peers involved in a Head.
  • The Hydra signing key for our node or the Hydra verification keys for peers do not match what's expected by each node. Check that AckSn messages are received by all parties and that the LogicOutcome log does not contain any Error