hidden cli args
hidden cli args are arguments that don't appear in the standard --help output but can still be used in the cli. these are often meant for advanced users, debugging, or experimental features.
- accounts db ancient append vecsAppendVecs that are older than (slots_per_epoch - SLOT-OFFSET) are squashed together."
- accounts db ancient storage ideal sizeThe smallest size of ideal ancient storage.
- accounts db create ancient storage packedCreate ancient storages in one shot instead of appending.
- accounts db experimental accumulator hashEnables the experimental accumulator hash
- accounts db hash calculation pubkey binsThe number of pubkey bins used for accounts hash calculation.
- accounts db max ancient storagesThe number of ancient storages the ancient slot combining should converge to.
- accounts db read cache limit mbHow large the read cache for account data can become, in mebibytes
- accounts db scan filter for shrinkingDebug option to use different type of filtering for accounts index scan in shrinking. "all" will scan both in-memory and on-disk accounts index, which is the default. "only-abnormal" will scan in-memory accounts index only for abnormal entries and skip scanning on-disk accounts index by assuming that on-disk accounts index contains only normal accounts index entry. "only-abnormal-with-verify" is similar to "only-abnormal", which will scan in-memory index for abnormal entries, but will also verify that on-disk account entries are indeed normal."
- accounts db snapshots use experimental accumulator hashSnapshots use the experimental accumulator hash
- accounts db squash storages methodSquash multiple account storage files together using this method
- accounts db test skip rewritesDebug option to skip rewrites for rent-exempt accounts but still add them in bank delta hash calculation"
- accounts db verify experimental accumulator hashVerifies the experimental accumulator hash
- accounts db verify refcountsDebug option to scan all append vecs and verify account index refcounts prior to clean"
- accounts filler countHow many accounts to add to stress the system. Accounts are ignored in operations related to correctness.
- accounts filler sizeSize per filler account in bytes.
- allow private addrAllow contacting private ip addresses
- delay leader block for pending forkDelay leader block creation while replaying a block which descends from the current fork and has a lower slot than our next leader slot. If we don't delay here, our new leader block will be on a different fork from the block we are replaying and there is a high chance that the cluster will confirm that block's fork rather than our leader block's fork because it was created before we started creating ours."
- no os cpu stats reportingDisable reporting of OS CPU statistics.
- no os disk stats reportingDisable reporting of OS disk statistics.
- no os memory stats reportingDisable reporting of OS memory statistics.
- no os network limits testSkip checks for OS network limits.
- no os network stats reportingDisable reporting of OS network statistics.
- no poh speed testSkip the check for PoH speed.
- no port checkDo not perform TCPUDP reachable port checks at start-up
- no skip initial accounts db cleanDo not skip the initial cleaning of accounts when verifying snapshot bank
- no wait for vote to start leaderIf the validator starts up with no ledger, it will wait to start block production until it sees a vote land in a rooted slot. This prevents double signing. Turn off to risk double signing a block."
- num quic endpointsThe number of QUIC endpoints used for TPU and TPU-Forward. It can be increased to increase network ingest throughput, at the expense of higher CPU and general validator load.
- partitioned epoch rewards compare calculationDo normal epoch rewards distribution, but also calculate rewards using the partitioned rewards code path and compare the resulting vote and stake accounts"
- partitioned epoch rewards force enable single slotForce the partitioned rewards distribution, but distribute all rewards in the first slot in the epoch. This should match consensus with the normal rewards distribution."
- poh hashes per batchSpecify hashes per batch in PoH service
- poh pinned cpu coreEXPERIMENTAL: Specify which CPU core PoH is pinned to
- process ledger before servicesProcess the local ledger fully before starting networking services
- repair whitelistA list of validators to prioritize repairs from. If specified, repair requests from validators in the list will be prioritized over requests from other validators. [default: all validators]"
- rocksdb ledger compressionThe compression algorithm that is used to compress transaction status data. Turning on compression can save ~10% of the ledger size."
- rocksdb perf sample intervalControls how often RocksDB readwrite performance samples are collected. Perf samples are collected in 1 ROCKS_PERF_SAMPLE_INTERVAL sampling rate."
- rpc bigtable instanceName of BigTable instance to target
- rpc send transaction batch sizeThe size of transactions to be sent in batch.
- rpc send transaction batch msThe rate at which transactions sent via rpc service are sent in batch.
- tpu max connections per ipaddr per minuteControls the rate of the clients connections per IpAddr per minute.
- tpu max connections per peerControls the max concurrent connections per IpAddr.
- tpu max fwd staked connectionsControls the max concurrent connections for TPU-forward from staked nodes.
- tpu max fwd unstaked connectionsControls the max concurrent connections for TPU-forward from unstaked nodes.
- tpu max staked connectionsControls the max concurrent connections for TPU from staked nodes.
- tpu max streams per msControls the max number of streams for a TPU service.
- tpu max unstaked connectionsControls the max concurrent connections fort TPU from unstaked nodes.
- tpu use quicUse QUIC to send transactions.
- vote use quicControls if to use QUIC to send votes.
- wen restartOnly used during coordinated cluster restarts. nn Need to also specify the leader's pubkey in --wen-restart-leader. nn When specified, the validator will enter Wen Restart mode which pauses normal activity. Validators in this mode will gossip their last vote to reach consensus on a safe restart slot and repair all blocks on the selected fork. The safe slot will be a descendant of the latest optimistically confirmed slot to ensure we do not roll back any optimistically confirmed slots. nn The progress in this mode will be saved in the file location provided. If consensus is reached, the validator will automatically exit with 200 status code. Then the operators are expected to restart the validator with --wait_for_supermajority and other arguments (including new shred_version, supermajority slot, and bankhash) given in the error log before the exit so the cluster will resume execution. The progress file will be kept around for future debugging. nn If wen_restart fails, refer to the progress file (in proto3 format) for further debugging and watch the discord channel for instructions."
- wen restart coordinatorSpecifies the pubkey of the leader used in wen restart. May get stuck if the leader used is different from others."
- on x /
- github /
- web /
- [email protected]