beacon-chain

The beacon-chain binary runs the Prysm beacon node. For the flags most people set, see Common flags and config file.

Generated from Prysm v7.3.1. Each flag name is a permalink you can share. Press / to filter.

General

Accepts Terms and Conditions (for non-interactive environments).

Without this flag the node prompts you to type accept on first start, which fails in systemd, Docker and other non-interactive environments. Once accepted (by flag or prompt), Prysm writes an empty tosaccepted file into --datadir and never asks again for that directory.

--accept-terms-of-use

Beacon node and APIs

--api-timeout=<duration>

Specifies the timeout value for API requests in seconds.

Wraps every Beacon API (HTTP) request in a timeout, except the /eth/v1/events stream. By default there is no timeout. The value is a Go duration, so include a unit (30s, 2m); 0 disables the timeout.

--api-timeout=30s

Peer-to-peer networking

--bootstrap-node=<string list>

The enr/enode address of bootstrap node. Beacon node will connect for peer discovery via DHT. Multiple nodes can be passed by using the flag multiple times but not comma-separated. You can also pass YAML files containing multiple nodes.

Sets the discv5 bootnodes used to find the first peers. The default shown here is the mainnet list; --hoodi and --sepolia load their own lists automatically, so you only need this flag on custom networks or to use your own bootnodes. Passing it replaces the whole built-in list rather than adding to it. Only enr: and enode: records are used (multiaddrs are dropped), and a value ending in .yaml is read as a YAML list of records.

Because the flag replaces the defaults, a single wrong or unreachable ENR leaves the node with no way to discover peers.

--bootstrap-node=enr:-Ku4Q... --bootstrap-node=/path/to/bootnodes.yaml

Database and storage

--backfill-batch-size=<uint64>

Number of blocks per backfill batch. A larger number will request more blocks at once from peers, but also consume more system memory to hold batches in memory during processing. This has a multiplicative effect with backfill-worker-count.

Blocks requested from a peer per backfill batch. Memory use grows with batch size × worker count × block size, so raise it only if backfill is slow and you have RAM to spare.

--backfill-batch-size=64

uint64default 32

--backfill-worker-count=<int>

Number of concurrent backfill batch requests. A larger number will better utilize network resources, up to a system-dependent limit, but will also consume more system memory to hold batches in memory during processing. Multiply by backfill-batch-size and average block size (~2MB before deneb) to find the right number for your system. This has a multiplicative effect with backfill-batch-size.

Number of backfill batches downloaded in parallel. More workers use more bandwidth and memory; the effect multiplies with --backfill-batch-size.

--backfill-worker-count=4

intdefault 2

--enable-backfill

Backfill will only be enabled when this flag is specified on a node using checkpoint sync.

After a checkpoint sync, downloads the blocks older than the checkpoint (and the blobs and data columns still inside their retention window) in the background, so the node can serve them to peers. By default it goes back MIN_EPOCHS_FOR_BLOCK_REQUESTS (33024 epochs, about 5 months on mainnet); it does nothing on a node synced from genesis. Deprecated alias: --enable-experimental-backfill.

booleandefault false

--clear-db

Prompt for clearing any previously stored data at the data directory.

Asks for confirmation at startup, then deletes the beacon database, the saved genesis state file, blob storage, data column storage and (if --slasher is set) the slasher database, and starts syncing from scratch. Files in the backups directory are kept. Remove the flag after the run you needed it for.

booleandefault false

--datadir=<string>

Data directory for the databases.

Root of everything the beacon node stores: the database in beaconchaindata/beaconchain.db, the genesis state file (genesis-<time>-<root>.ssz), and by default the blobs/ and data-columns/ directories. The default depends on the OS: ~/.eth2 on Linux, ~/Library/Eth2 on macOS, %LOCALAPPDATA%\Eth2 on Windows.

Use a separate --datadir per network. Prysm refuses a database whose stored deposit contract differs from the selected network's (database contract is … but tried to run with …), but mainnet and Hoodi use the same address, so that check doesn't catch a mix-up between those two.

--datadir=/var/lib/ethereum/beacon

stringdefault ~/Library/Eth2

--force-clear-db

Clears any previously stored data at the data directory.

Same as --clear-db, but deletes without asking. Use it in non-interactive setups (systemd, Docker) where there is no terminal to answer the prompt.

If you leave this flag in a service definition, the node wipes its database and resyncs on every restart.

booleandefault false

--restore-source-file=<string>

Filepath to the backed-up database file which will be used to restore the database

Only read by the beacon-chain db restore subcommand; setting it on a normal beacon node run does nothing. Point it at a copy of a beaconchain.db file.

string

--restore-target-dir=<string>

Target directory of the restored database

Only read by the beacon-chain db restore subcommand; setting it on a normal beacon node run does nothing. Use the same value you pass to --datadir.

stringdefault ~/Library/Eth2

--backfill-oldest-slot=<uint64>

Specifies the oldest slot that backfill should download. If this value is greater than current_slot - MIN_EPOCHS_FOR_BLOCK_REQUESTS, it will be ignored with a warning log.

Makes backfill go further back than the spec minimum, down to this slot (for example 1 for the full history). A value newer than current slot - MIN_EPOCHS_FOR_BLOCK_REQUESTS is ignored, with the warning Ignoring user-specified slot > MIN_EPOCHS_FOR_BLOCK_REQUESTS.

--backfill-oldest-slot=1

uint64default 0

--blob-retention-epochs=<uint64>

Override the default blob retention period (measured in epochs). The node will exit with an error at startup if the value is less than the default of 4096 epochs.

How many epochs of blobs and data columns the node keeps before deleting them. The same value applies to both; the minimum is the spec value of 4096 epochs (about 18 days), and a lower value stops the node at startup. Raise it if you run an indexer or L2 tooling that needs older blob data. Deprecated alias: --extend-blob-retention-epoch.

--blob-retention-epochs=8192

uint64default 4096

--beacon-db-pruning

Enables pruning of beacon db beyond MIN_EPOCHS_FOR_BLOCK_REQUESTS duration. This is an opt-in feature, and should only be enabled if operators doesn't require historical data.

Starts a pruner that deletes blocks and states older than the retention window from the database, once per epoch (at mid-epoch), after initial sync and backfill have finished. Blobs and data columns are always pruned by their own retention setting, with or without this flag.

Pruned history is gone: the node can no longer answer API queries for blocks or states before the window. Don't enable it on a node that serves historical data.

booleandefault false

--pruner-retention-epochs=<uint64>

Specifies the retention period for the pruner service in terms of epochs. If this value is less than MIN_EPOCHS_FOR_BLOCK_REQUESTS, it will be ignored.

How many epochs of blocks and states the pruner keeps. Only used with --beacon-db-pruning. The minimum is MIN_EPOCHS_FOR_BLOCK_REQUESTS + 1 (33025 epochs on mainnet); lower values are replaced by the minimum with the warning Retention period too low, ignoring and using minimum required value.

--pruner-retention-epochs=50000

uint64default 0

--slots-per-archive-point=<int>

The slot durations of when an archived state gets saved in the beaconDB.

How often a full finalized state is saved to the database (default every 2048 slots, 64 epochs). Any other historical state is regenerated by replaying blocks from the closest saved one, so a lower value makes historical state queries faster at the cost of much more disk. Not used when --enable-state-diff is on, which stores history differently.

An explicit value also overrides the 128 slots set by --historical-slasher-node.

--slots-per-archive-point=256

intdefault 2048

--blob-storage-layout=<string>

Dictates how to organize the blob directory structure on disk, available options are: flat, by-epoch.

Only affects blob sidecars, which are stored for the Deneb and Electra forks; from Fulu on, data column sidecars are stored separately and this flag doesn't apply to them. by-epoch groups blob files in per-epoch directories, which makes startup and pruning faster; flat is the old one-directory-per-block layout. Without the flag, Prysm detects the layout already on disk and warns if it is flat; set --blob-storage-layout=by-epoch once to migrate it on the next restart.

--blob-storage-layout=by-epoch

Block building and MEV

--local-block-value-boost=<uint64>

A percentage boost for local block construction as a Uint64. This is used to prioritize local block construction over relay/builder block constructionBoost is an additional percentage to multiple local block value. Use builder block if: builder_bid_value * 100 > local_block_value * (local-block-value-boost + 100)

Handicaps builder bids against your locally built block. The beacon node picks the builder block only if builder_bid_gwei * builder_boost_factor > local_value_gwei * (100 + local-block-value-boost), where builder_boost_factor is the per-request value sent by the validator client (default 100). The check runs after --min-builder-bid and --min-builder-to-local-difference, and applies to blocks up to the Fulu fork.

Set it to 0 if you want the builder block whenever it pays more, however slightly. If you also send a non-default builder_boost_factor from the validator client, both boosts apply and the node logs a warning.

--local-block-value-boost=20

uint64default 10

--max-builder-consecutive-missed-slots=<int>

Number of consecutive skip slot to fallback from using relay/builder to local execution engine for block construction

First circuit breaker: if the gap between the proposal slot and the highest slot the node has received a block for reaches this number, the node stops asking the relay and builds the block locally. It protects chain liveness when relays fail. The node logs Circuit breaker activated due to missing consecutive slot when it trips.

--max-builder-consecutive-missed-slots=3

intdefault 3

--max-builder-epoch-missed-slots=<int>

Number of total skip slot to fallback from using relay/builder to local execution engine for block construction in last epoch rolling window

Second circuit breaker: if the number of empty slots in the last 32 slots (a rolling epoch window) reaches this number, the node builds locally instead of using the relay. It logs Circuit breaker activated due to missing enough slots last epoch when it trips.

--max-builder-epoch-missed-slots=5

intdefault 5

--http-mev-relay=<string>

A MEV builder relay string http endpoint, this will be used to interact MEV builder network using API defined in: https://ethereum.github.io/builder-specs/#/Builder

Point this at MEV-Boost (or another Builder API endpoint), not at an individual relay. With it set, the node asks the builder for a bid on proposals from validators that registered with the builder (validator client --enable-builder or the builder section of the proposer settings file); other proposals, and proposals where the circuit breaker is tripped, use the local execution client.

--http-mev-relay=http://127.0.0.1:18550

string

--min-builder-bid=<uint64>

An absolute value in Gwei that the builder bid has to have in order for this beacon node to use the builder's block. Anything less than this value and the beacon will revert to local building.

Floor for builder bids, in Gwei. If a bid is below this value the node uses the local payload and logs Proposer: using local execution payload because min bid not attained. 1 ETH is 1000000000 Gwei. Applies to blocks up to the Fulu fork.

--min-builder-bid=10000000

uint64default 0

--min-builder-to-local-difference=<uint64>

An absolute value in Gwei that the builder bid has to have in order for this beacon node to use the builder's block. Anything less than this value and the beacon will revert to local building.

When the builder bid is below the local block value plus this many Gwei, the node builds locally and logs Proposer: using local execution payload because min difference with local value was not attained. It is checked after --min-builder-bid and before --local-block-value-boost. Applies to blocks up to the Fulu fork.

--min-builder-to-local-difference=5000000

uint64default 0

--builder-header-timeout=<duration>

Timeout to use when fetching a block header from the builder API, as a duration (e.g. 1s, 2s, 2500ms). Must be greater than 0. Only effective up to the Fulu fork.

How long the node waits for the getHeader response from MEV-Boost before giving up and proposing the local block. Accepts Go durations (950ms, 1s). Zero or negative values stop the node at startup, and any override logs a warning at startup.

Raising it delays your block. The default already matches the builder spec's proposal delay tolerance of 1 second; a much higher value risks a missed or orphaned proposal.

--builder-header-timeout=1s

durationdefault 1s

--builder-bid-timeout=<duration>

Timeout to use when fetching execution payload bids from the builder API, as a duration (e.g. 600ms, 1s). Must be greater than 0. Only effective from the Gloas fork onward.

From the Gloas fork on, how long the node waits for execution payload bids from builders when it builds a block. Accepts Go durations (600ms, 1s). Must be greater than 0. Has no effect before Gloas; --builder-header-timeout plays that role for MEV-Boost.

Raising it delays your block, which risks a missed or orphaned proposal.

--builder-bid-timeout=600ms

Sync

--checkpoint-block=<path>

Rather than syncing from genesis, you can start processing from a ssz-serialized BeaconState+Block. This flag allows you to specify a local file containing the checkpoint Block to load.

The block matching --checkpoint-state, as a local SSZ file. Both files must be given together, and the block must be the one the state was built from.

--checkpoint-block=/data/block_mainnet_fulu_12345678-0x….ssz

path

--checkpoint-sync-url=<string>

URL of a synced beacon node to trust in obtaining checkpoint sync data. As an additional safety measure, it is strongly recommended to only use this option in conjunction with --weak-subjectivity-checkpoint flag

Downloads the latest finalized state and its block from this Beacon API (/eth/v2/debug/beacon/states/finalized) and starts from there, so the node is synced in minutes instead of days. It is only used on an empty database: if one already exists, the log shows Origin checkpoint found in the database, ignoring checkpoint sync flags. When --genesis-beacon-api-url is not set, the same URL is also used to fetch the genesis state.

You trust this endpoint to give you the canonical chain. Use a provider you trust, and compare the finalized block root with a second source.

--checkpoint-sync-url=https://hoodi.beaconstate.ethstaker.cc

string

--checkpoint-state=<path>

Rather than syncing from genesis, you can start processing from a ssz-serialized BeaconState+Block. This flag allows you to specify a local file containing the checkpoint BeaconState to load.

Starts from a finalized state stored in a local SSZ file instead of a URL, for example one saved with prysmctl checkpoint-sync download. It needs --checkpoint-block too; the node refuses to start with only one of the two. If --checkpoint-sync-url is also set, the URL wins and the files are ignored.

--checkpoint-state=/data/state_mainnet_fulu_12345678-0x….ssz

path

--weak-subjectivity-checkpoint=<string>

Input in `block_root:epoch_number` format. This guarantees that syncing leads to the given Weak Subjectivity Checkpoint along the canonical chain. If such a sync is not possible, the node will treat it as a critical and irrecoverable failure

A block root and epoch the chain must pass through. Once the node finalizes past that epoch, it checks that the block is in its database and finalized, and logs Weak subjectivity check has passed!!; otherwise it exits with Could not verify weak subjectivity checkpoint. Use it with checkpoint sync to detect a dishonest checkpoint provider. A 0x prefix on the root is accepted.

--weak-subjectivity-checkpoint=0x<BLOCK_ROOT>:<EPOCH>

string

--genesis-beacon-api-url=<string>

URL of a synced beacon node to trust for obtaining genesis state. As an additional safety measure, it is strongly recommended to only use this option in conjunction with --weak-subjectivity-checkpoint flag

Fetches the genesis state from this Beacon API (the debug state endpoint for genesis) when there is none in the data directory. You rarely need it: it defaults to the value of --checkpoint-sync-url, and mainnet's genesis is embedded. Set it for a testnet you sync from genesis without a local genesis file.

--genesis-beacon-api-url=https://hoodi.beaconstate.ethstaker.cc

string

--genesis-state=<path>

Load a genesis state from ssz file. Testnet genesis files can be found in the eth2-clients/eth2-testnets repository on github.

Loads the genesis state from a local SSZ file. Only mainnet's genesis state is built into Prysm; other networks get it from this flag, --genesis-beacon-api-url or --checkpoint-sync-url. It is only needed once: the state is saved in the data directory and reused. This flag takes priority over --genesis-beacon-api-url.

--genesis-state=/data/hoodi/genesis.ssz

path

Execution client connection

--contract-deployment-block=<int>

The eth1 block in which the deposit contract was deployed.

Execution block number where Prysm starts reading deposit contract logs. The default shown is mainnet's; each network flag sets its own value (0 for Hoodi, 1273020 for Sepolia). Only needed for custom networks, and only until EIP-6110 deposit requests are active.

--contract-deployment-block=0

intdefault 11184524

--deposit-contract=<string>

Deposit contract address. Beacon chain node will listen logs coming from the deposit contract to determine when validator is eligible to participate.

Overrides the deposit contract address of the selected network. The default shown is mainnet's; each network flag sets the right address for its network. The address is saved in the database, and the node refuses to start if it later differs (mainnet and Hoodi share the same address, so this doesn't tell those two apart). Only needed for custom networks.

--deposit-contract=0x0000000000000000000000000000000000000000

stringdefault 0x00000000219ab540356cBB839Cbe05303d7705Fa

--engine-endpoint-timeout-seconds=<uint64>

Sets the execution engine timeout (seconds) for execution payload semantics (forkchoiceUpdated, newPayload)

Timeout, in seconds, for engine_newPayload and engine_forkchoiceUpdated calls. When not set, Prysm uses 8 seconds (the Engine API spec value); the 0 shown as default means "not set". Raise it only if a slow execution client times out while importing blocks.

Explicitly setting 0 means a zero timeout, so every call fails.

--engine-endpoint-timeout-seconds=12

uint64default 0

--eth1-header-req-limit=<uint64>

Sets the maximum number of headers that a deposit log query can fetch.

Maximum block range per request when reading deposit contract logs and headers from the execution client; Prysm shrinks the range automatically when requests fail. Lower it if your execution client or provider rejects large log queries. It only matters until the chain uses EIP-6110 deposit requests (Electra): after that, Prysm stops reading deposit logs.

--eth1-header-req-limit=150

uint64default 1000

--execution-endpoint=<string>

An execution client http endpoint. Can contain auth header as well in the format

Engine API endpoint of your execution client, normally its authenticated port 8551. Also accepts an IPC socket path (a plain path or ipc://), which needs no JWT. If the endpoint is down, Prysm keeps retrying and logs Connected to new endpoint once it reconnects. An Authorization suffix after a comma (<URL>,Basic <USER:PASSWORD>) is a legacy option: with --jwt-secret set, Prysm replaces it with JWT authentication.

--execution-endpoint=http://localhost:8551

stringdefault http://localhost:8551

--execution-headers=<string>

A comma separated list of key value pairs to pass as HTTP headers for all execution client calls. Example: --execution-headers=key1=value1,key2=value2

Extra HTTP headers sent with every call to the execution client, for example an API key for an authenticating proxy or hosted endpoint. Pairs are split on , and each key on the first =, so values can contain = but not commas. A pair without = is skipped with a warning. An Authorization header set here is overwritten by the JWT on every request when --jwt-secret is set.

--execution-headers=X-Api-Key=<API_KEY>

string

--jwt-secret=<string>

REQUIRED if connecting to an execution node via HTTP. Provides a path to a file containing a hex-encoded string representing a 32 byte secret used for authentication with an execution node via HTTP. If this is not set, all requests to execution nodes via HTTP for consensus-related calls will fail, which will prevent your validators from performing their duties. This is not required if using an IPC connection.

Path to the file with the shared secret the execution client also uses: 32 bytes as hex, with or without 0x. Whitespace around it is ignored; an empty, unreadable or wrong-length file stops the node at startup. On success it logs Finished reading JWT secret from <path>. Create one with beacon-chain generate-auth-secret or openssl rand -hex 32.

Without it, HTTP calls to the execution client get 401 Unauthorized, and your validators can't do their duties.

--jwt-secret=/var/lib/ethereum/jwt.hex

string

--jwt-id=<string>

JWT claims id. Could be used to identify the client

Sets the optional id claim in every JWT sent to the execution client. It doesn't change authentication; use it to tell several consensus clients apart in the execution client's logs.

--jwt-id=prysm-node-1

Monitoring

--disable-monitoring

Disables monitoring service.

Turns off the Prometheus HTTP server on --monitoring-host:--monitoring-port. That also removes /healthz, /goroutinez and /p2p, which are served on the same port, so don't use it if a health check or dashboard depends on them.

booleandefault false

--enable-tracing

Enables OpenTelemetry tracing.

Starts an OpenTelemetry tracer that exports spans over OTLP/HTTP to --tracing-endpoint with service name beacon-chain. It adds CPU and network overhead, so enable it for debugging rather than permanently.

--enable-tracing

booleandefault false

--monitoring-host=<string>

Host used for listening and responding metrics for prometheus.

Interface the metrics server binds to. The same server also serves /healthz, /goroutinez and /p2p. Set 0.0.0.0 only if Prometheus runs on another machine or in another container, and firewall the port.

--monitoring-host=0.0.0.0

stringdefault 127.0.0.1

--trace-sample-fraction=<float64>

Head-sampling ratio applied to every trace started by the process. 1.0 keeps all traces.

Fraction of traces kept, from 0 to 1. It is a head-based ratio sampler on the trace ID, so a trace is kept or dropped as a whole. Lower it on busy nodes to reduce overhead; only used with --enable-tracing.

--trace-sample-fraction=0.05

float64default 0.2

--tracing-endpoint=<string>

OTLP/HTTP endpoint that traces are exported to (e.g. an OpenTelemetry Collector or a Tempo receiver).

Full URL, including the path, that spans are posted to with the OTLP/HTTP exporter. The default points at the old Jaeger collector path; for an OpenTelemetry Collector, Tempo or recent Jaeger, use the OTLP/HTTP receiver, usually port 4318 with path /v1/traces.

--tracing-endpoint=http://127.0.0.1:4318/v1/traces

stringdefault http://127.0.0.1:14268/api/traces

--tracing-process-name=<string>

Name to apply to tracing tag `process_name`.

Adds a process_name resource attribute (and sets service.instance.id) on every exported span, so you can tell several beacon nodes apart in the same tracing backend.

--tracing-process-name=beacon-hoodi-1

string

--monitor-indices=<int list>

List of validator indices to track performance

Starts the validator monitor for these validator indices (not public keys). After initial sync the node logs events such as Proposed beacon block was included, Attestation included and a periodic Aggregated performance since launch, and exports monitor_* Prometheus metrics labeled by validator_index. Pass a comma-separated list or repeat the flag.

--monitor-indices=1234,5678

int list

--monitoring-port=<int>

Port used to listening and respond metrics for Prometheus.

Port of the metrics server. Prometheus scrapes /metrics here; the same port serves /healthz (HTTP 503 if any service reports an error), /goroutinez (goroutine dump) and /p2p (peer summary). If the port is taken, the node keeps running without metrics and logs Port already in use; cannot start prometheus service.

--monitoring-port=8080

intdefault 8080

Slasher

--historical-slasher-node

Enables required flags for serving historical data to a slasher client. Results in additional storage usage

Prepares a beacon node to feed historical data to a slasher: it saves a full state every 4 epochs (128 slots) instead of every 2048 slots, and raises the maximum API page size to 4096 items. This uses much more disk.

An explicit --slots-per-archive-point overrides the 128-slot setting.

booleandefault false

--slasher-datadir=<string>

Directory for the slasher database

Base directory for the slasher database, stored at <dir>/beaconchaindata/slasher.db. Defaults to --datadir, so the slasher database sits next to the beacon database; set it to put the slasher on another disk.

--slasher-datadir=/mnt/slasher

stringdefault ~/Library/Eth2

--slasher

Enables a slasher in the beacon node for detecting slashable offenses.

Runs a slasher inside the beacon node: it watches attestations and blocks from gossip, detects double votes, surround votes and double proposals, and adds the slashings to the pool so they get included in blocks. It starts detecting only after the node has synced (Completed chain sync, starting slashing detection), and needs extra disk, memory and CPU. It is not required to protect your own validators; slashing protection lives in the validator client.

booleandefault false

Logging

--disable-log-colors

Disable color formatting for terminal logs.

Removes ANSI color codes from terminal output in the text format. Useful when stdout goes to a file or log collector that shows escape codes. Files written with --log-file and the debug log file never contain colors.

booleandefault false

--log-format=<string>

Specifies log formatting. Supports: text, json, fluentd, journald.

text (default) is the human-readable format and the only one that honors --log.vmodule. json and fluentd write one JSON object per line to stderr for log shippers. journald sends entries straight to the systemd journal; it works only on Linux, and if the journal isn't reachable the node logs a warning and ignores the setting.

--log-format=json

stringdefault text

--log-file=<string>

Specifies log file name, relative or absolute.

Also writes logs to this file, appending on restart and creating parent directories as needed. With --log-format=text the file gets uncolored Prysm text logs that honor --log.vmodule; with other formats the file is written in logrus's plain key=value format rather than JSON. Prysm does not rotate this file, so use logrotate or similar.

--log-file=/var/log/prysm/beacon.log

string

--verbosity=<string>

Logging verbosity. (trace, debug, info, warn, error, fatal, panic)

Minimum level printed to the terminal and to --log-file. debug is useful for troubleshooting but very noisy; trace is for developers. Independently of this flag, the node keeps a debug-level log file in <datadir>/logs/ unless you pass --disable-ephemeral-log-file.

--verbosity=debug

stringdefault info

--disable-ephemeral-log-file

Disables the creation of a debug log file that keeps 24 hours of logs.

By default the node writes a debug-level, uncolored copy of its logs to <datadir>/logs/beacon-chain.log, whatever --verbosity is set to. The file rotates at 250 MB, keeps one backup, and removes files older than one day. Disable it to save disk writes, at the cost of losing the debug history support usually asks for.

booleandefault false

--log.vmodule=<string list>

Per-package log verbosity. packagePath=level entries separated by commas.

Overrides --verbosity for specific packages. Keys are Prysm package paths without the module prefix, as shown in the package field of JSON logs (for example beacon-chain/sync). An entry also covers its subpackages, and the longest match wins. Wildcards, leading or trailing slashes and duplicates are rejected at startup. Only the text log format honors it.

--log.vmodule=beacon-chain/sync=debug,beacon-chain/p2p=trace

string list

Feature flags

--save-invalid-block-temp

Writes invalid blocks to temp directory.

When a gossiped block fails validation, Prysm writes it as SSZ to the operating system's temp directory ($TMPDIR or /tmp on Linux) as beacon_block_<slot>.ssz and logs Writing invalid block to disk at <path>. Use it to capture a block for a bug report; it is safe in production but files accumulate until you delete them.

deprecated

--dev

Enables experimental features still in development. These features may not be stable.

A shortcut that turns on the feature flags Prysm's developers want tested before they become defaults. In this release it only sets --enable-backfill (unless you set that flag yourself) and logs Enabling development mode flags at startup. The set changes between releases, so don't use it in production; enable the individual flags you want instead. Deprecated: it still works, but will be removed.

Debugging and profiling

--healthz-max-goroutines=<int>

The /healthz endpoint reports the node as unhealthy if the average goroutine count during a 10 slots window exceeds this threshold.

Threshold for the blockchain service's health status. If the average goroutine count exceeds it, the service reports an error and /healthz on the monitoring port returns HTTP 503. Raise it if your node is healthy but trips this check under heavy load. --max-goroutines is a deprecated alias.

--healthz-max-goroutines=10000

intdefault 5000

--blockprofilerate=<int>

Turns on block profiling with the given rate.

Sets Go's runtime.SetBlockProfileRate: about one blocking event is sampled per this many nanoseconds spent blocked (1 records every event). 0 (default) leaves block profiling off. Read results from /debug/pprof/block with --pprof.

--blockprofilerate=1

intdefault 0

--memprofilerate=<int>

Turns on memory profiling with the given rate.

Sets Go's runtime.MemProfileRate: one allocation sampled per this many bytes allocated. The default 524288 (512 KiB) is Go's own default. Lower values give more precise heap profiles at a CPU cost; 0 turns heap profiling off.

Subcommands

`beacon-chain db restore`

restores a database from a backup file

--restore-source-file=<altsrc.string>

Filepath to the backed-up database file which will be used to restore the database

Path to the database file to restore. The file is copied as is to <restore-target-dir>/beaconchaindata/beaconchain.db; blobs and data columns are not part of it. Stop the beacon node before restoring.

beacon-chain db restore --restore-source-file=/backups/beaconchain.db --restore-target-dir=/var/lib/ethereum/beacon

Read the guide

--restore-target-dir=<altsrc.string>

Target directory of the restored database

The beacon node data directory to restore into, the same value as --datadir. If a database already exists there, the command asks before overwriting it.

Read the guide

default ~/Library/Eth2

`beacon-chain generate-auth-secret`

creates a random, 32 byte hex string in a plaintext file to be used for authenticating JSON-RPC requests. If no --output-file flag is defined, the file will be created in the current working directory

--output-file=<altsrc.string>: Target file path for outputting a generated JWT secret to be used for JSON-RPC authentication
Where to write the new JWT secret: a 0x-prefixed 64-character hex string (32 random bytes), with 0600 permissions. Without this flag the file is jwt.hex in the current directory; missing parent directories are created. Give the same file to the beacon node (--jwt-secret) and the execution client.
An existing file at that path is overwritten without asking. Restart both clients if the secret changes.
beacon-chain generate-auth-secret --output-file=/var/lib/ethereum/jwt.hex
Read the guide