Introduction
This document describes how to configure network monitoring within the node provided by aemon application.
Configuration
By default monitoring is disabled. To turn it on monitoring.publisher.pubkey and monitoring.active = true need to be setup. To start posting transaction setup monitoring.publisher.autostart to true.
To post transactions account's public and private key is needed. For online processing only public key needs to be configured. Offline post-processing can be done by filtering transaction by publisher public key.
Metrics configuration
Monitoring uses statsd backend provided by apps/aecore/src/aec_metrics.erl. See metrics metrics.* configuration keys in apps/aeutils/priv/aeternity_config_schema.json
Metrics
Each metric uses ae.epoch.aemon. prefix.
| Name | Type | Description |
|---|---|---|
confirmation.delay |
histogram | Number of keyblock created before signing transaction |
forks.micro.count |
counter | Count of observed micro-forks |
forks.micro.height |
histogram | Height difference of observed micro-forks |
gen_stats.microblocks.total |
histogram | Number of microblocks in a generation |
gen_stats.tx.monitoring |
histogram | Number of monitoring transactions in a generation |
gen_stats.tx.total |
histogram | Number of transactions in a generation |
publisher.balance |
gauge | Publisher balance |
publisher.post_tx.max_adjustment |
counter | Transaction posting error: |
publisher.post_tx.nonce_too_high |
counter | Transaction posting error: |
publisher.post_tx.nonce_too_low |
counter | Transaction posting error: |
publisher.post_tx.success |
counter | Successful transaction posts |
publisher.queue.size |
histogram | Number of transactions posted but not signed on chain |
publisher.queue.ttl_expired |
histogram | Number of transactions with expired ttl |
block.propagation_time.key |
histogram | Time until key-blocks reached this node in milliseconds |
block.propagation_time.micro |
histogram | Time until micro-blocks reached this node in milliseconds |
block.time_since_prev.key |
histogram | Time between key-blocks in milliseconds |
block.time_since_prev.micro |
histogram | Time between micro-blocks in milliseconds |
block.tx.total.micro |
histogram | Number of transactions in a microblock |
block.gas.total.micro |
histogram | Gas used per microblock |
block.gas.per_tx.micro |
histogram | Gas used per transaction in a microblock |
block.size.per_tx.micro |
histogram | Size of transactions in a microblock in bytes |
chain.top.difficulty |
gauge | Difficulty of the top block |
How to read metrics
confirmation.delay
represents network latency. A high number might imply a busy network or unfair leaders.
forks.micro & gen_stats.microblocks.total
forks.micro represents the length of microfork. It shows a minimum number observed by monitoring, not an exact one. Based on behaviour observed in mainnet in the first half of 2019, roughly 33% of microblocks transactions are rewritten to the next generation. Use gen_stats.microblocks.total as a reference.
gen_stats.microblocks.total and gen_stats.tx.{monitoring,total}
Statistic metrics can be used to measure network saturation
publisher.post_tx.*
Metrics can be used to monitor mempool's transaction propagation. When publisher.post_tx.nonce_too_high is preset you might want to check mempool.nonce_offset configuration/
publisher.queue.*
For further network transaction propagation investigation. All transactions accepted by mempool are tracked by *.size. Over time it should correlate with gen_stats.tx.monitoring.
*.ttl_expired might imply low transaction fee or busy network.