Directory / cardano-sl / configuration.md

You are browsing a mirror of a file hosted on GitHub. View original

Cardano SL configuration

This document describes Cardano SL configuration. Note: the description is valid for git revision b674421cc. It may be outdated for newer version of the code.

Table of contents

Prerequisites

  • Read about VSS certificates.
  • You need to know the difference between balance and stake.
  • You need to know about AVVM. In short: there was vending process after which we have a map from AVVM address (32 bytes, base64url encoded) to an integer number (amount of ADA). AVVM addresses can be converted to Cardano-SL addresses and initially each address has this amount of ADA (as balance, not stake).
  • Bootstrap era and stake locking.
  • SSC is Shared Seed Computation, it’s performed by rich nodes (with big stake) to agree upon the seed which will be used for leaders selection for an epoch.

Overview

Configuration is stored in YAML format, conventional name for the file is configuration.yaml. This file contains multiple configurations, each one is associated with a key. An example of configuration can be found in the file lib/configuration.yaml.

Configuration consists of several parts (for different parts of Cardano SL), each will be descirbed below:

  • core configuration

    • genesis configuration is the most important part of core, described

in a separate section

  • infra configuration
  • update configuration
  • ssc configuration
  • delegation configuration
  • node configuration

Command line options

Almost all executables accept several options to specify configuration.

  • --configuration-file can be used to specify path to configuration file. Default value is lib/configuration.yaml.
  • --configuration-key specifies key in this configuration. Default value is default.
  • --system-start specifies system start time as unix timestamp in seconds. It must be provided if it’s not available from configuration and must not be provided otherwise. More details are provided below.
  • --configuration-seed can be used to specify seed used to generate secret data. It overrides seed value from initializer (see below). If genesis data is used, passing --configuration-seed is prohibited and leads to a runtime error.

Genesis

In Cardano-SL we need to specify which addresses initially have ADA and how much. It’s traditionally called genesis block. In Cardano-SL we also have various parameters of the algorithm affecting consensus along with initial balances. The hash of this data (in canonical JSON format) is used as the parent of the very first real block.

Naming conventions

Traditionally all initial data is called genesis block. However, in Cardano-SL genesis block has a different meaning: it’s the first block in an epoch, as described in Ouroboros paper. There are many genesis blocks (one per epoch).

For this reason we call the set of all initial balances, stakes and parameters genesis data (and there is GenesisData type in Haskell code). Another term is genesis spec (data type in Haskell is called GenesisSpec) which specifies how to create genesis data. See below for more details. When we say simply genesis it usually means genesis data, the actual meaning is usually clear from the context.

Note: we are not trying to force everyone to change standards and say genesis data instead of genesis block (though it kinda makes sense, because in Cardano-SL genesis data is not a block, it consists of a bunch of different data). It’s ok to use term genesis block and usually the meaning will be clear from the context. But for historical reasons developers prefer to use term genesis data to refer to the initial set of parameters and balances and genesis block to refer to the first block in an epoch. It’s enforced by names in code. It may be changed later.

How to specify genesis

Configuration is used to specify genesis data used by the node. It is part of core configuration (it’s accessible by key <configuration-key>.core.genesis). There are two ways to specify genesis data.

  • One way is to provide genesis data itself, stored in JSON format. Genesis data explicitly states balances of all addresses, all bootstrap stakeholders, heavyweight delegation certificates, etc. It makes it clearer which genesis is used, but it doesn’t describe how this genesis was constructed. Also it requires more steps to create genesis data, than to create a spec. This mechanism is used for mainnet (both production and staging) so that everyone can open this file and see all balances (and everything else from genesis data) at any time. It also will be used for testnet, at least for the first one. The format is demonstrated below. hash is the hash (Blake2b 256 bits) of canonically encoded mainnet-genesis.json file. It can be computed manually using scripts/js/genesis-hash.js script.
    genesis:
      src:
        file: mainnet-genesis.json
        hash: 5f20df933584822601f9e3f8c024eb5eb252fe8cefb24d1317dc3d432e940ebb

The format of mainnet-genesis.json is described in Genesis data format.

  • Another way is to provide a specification of how to generate genesis data. It allows one to specify how many nodes should have stake, which bootstrap stakeholders should be used, what should be total balance, etc. When node is launched with spec as genesis, it automatically constructs genesis data from it. This way to specify genesis is supposed to be used for testing clusters. It’s questionable whether it should be used for testnet and currently we can’t use it for testnet for performance reasons. We don’t use it for mainnet. See Genesis spec format for details.

Genesis data format

Let’s start with an example:

{
    "avvmDistr": {
        "0cAZzmtB2CUjlsMULb30YcBrocvK7VhQjUk--MyY2_Q=": "8333333000000",
        ...
        "zE56EfVAvv0XekVyBDGh2Iz1QT6X38YxlcBYO20hqa0=": "1773135000000"
    },
    "blockVersionData": {
        "heavyDelThd": "300000000000",
        "maxBlockSize": "2000000",
        "maxHeaderSize": "2000000",
        "maxProposalSize": "700",
        "maxTxSize": "4096",
        "mpcThd": "20000000000000",
        "scriptVersion": 0,
        "slotDuration": "20000",
        "softforkRule": {
            "initThd": "900000000000000",
            "minThd": "600000000000000",
            "thdDecrement": "50000000000000"
        },
        "txFeePolicy": {
            "multiplier": "43946000000",
            "summand": "155381000000000"
        },
        "unlockStakeEpoch": "18446744073709551615",
        "updateImplicit": "10000",
        "updateProposalThd": "100000000000000",
        "updateVoteThd": "1000000000000"
    },
    "bootStakeholders": {
        "0d916567f96b6a65d204966e6aab5fbd242e56c321833f8ba5d607da": 1,
        "4bd1884f5ce2231be8623ecf5778a9112e26514205b39ff53529e735": 1,
        "5f53e01e1366aeda8811c2a630f0e037077a7b651093d2bdc4ef7200": 1,
        "ada3ab5c69b945c33c15ca110b444aa58906bf01fcfe55d8818d9c49": 1
    },
    "ftsSeed": "736b6f766f726f64612047677572646120626f726f64612070726f766f646120",
    "nonAvvmBalances": {},
    "protocolConsts": {
        "k": 2160,
        "protocolMagic": 60987900,
        "vssMaxTTL": 6,
        "vssMinTTL": 2
    },
    "heavyDelegation": {
        "c6c7fb227037a1719b9d871ea49b6039325aeb293915da7db9620e3f": {
            "cert": "0f2ff9d1f071793dbd90fce8d47e96a09b594ee6dd00a4bac9664e4bd6af89830035ec921698b774c779eb1b6a2772d3d6ae37e630c06c75fbfecd02a6410a02",
            "delegatePk": "I07+5HIUW0Lkq0poMMzGziuILwxSyTJ4lMSsoQz4HZunD5Xsx3MfBZWc4l+206lUOFaU+spdJg7MkmFKBoVV0g==",
            "issuerPk": "AZzlT+pC5M4xG+GuRHJEXS+isikmVBorTqOyjRDGUQ+Lst9fn1zQn5OGKSXK29G2dn7R7JCugfcUebr0Dq7wPw==",
            "omega": 1
        }
    },
    "startTime": 1505621332,
    "vssCerts": {
        "0d916567f96b6a65d204966e6aab5fbd242e56c321833f8ba5d607da": {
            "expiryEpoch": 1,
            "signature": "396fe505f4287f832fd26c1eba1843a49f3d23b64d664fb3c8a2f25c8de73ce6f2f4cf37ec7fa0fee7750d1d6c55e1b07e1018ce0c6443bacdb01fb8e15f1a0f",
            "signingKey": "ohsV3RtEFD1jeOzKwNulmMRhBG2RLdFxPbcSGbkmJ+xd/2cOALSDahPlydFRjd15sH0PkPE/zTvP4iN8wJr/hA==",
            "vssKey": "WCECtpe8B/5XPefEhgg7X5veUIYH/RRcvXbz6w7MIJBwWYU="
        },
        ...
    }
}

Section "avvmDistr" contains AVVM addresses with corresponding balances (lovelaces).

Section "blockVersionData" contains fundamental blockchain-related values:

  • "heavyDelThd" - heavyweight delegation threshold,
  • "maxBlockSize" - maximum size of block, in bytes,
  • "maxHeaderSize" - maximum size of block’s header, in bytes,
  • "maxProposalSize" - maximum size of Cardano SL update proposal, in bytes,
  • "maxTxSize" - maximum size of transaction, in bytes,
  • "mpcThd" - threshold for participation in SSC (shared seed computation),
  • "scriptVersion" - script version,
  • "slotDuration" - slot duration, in milliseconds,
  • "softforkRule" - rules for softfork:

    • "initThd" - initial threshold, right after proposal is confirmed,
    • "minThd" - minimal threshold (i.e. threshold can’t become less than this one),
    • "thdDecrement" - theshold will be decreased by this value after each epoch,
  • "txFeePolicy" - transaction fee policy’s values,
  • "unlockStakeEpoch" - unlock stake epoch after which bootstrap era ends,
  • "updateImplicit" - update implicit period, in slots,
  • "updateProposalThd" - threshold for Cardano SL update proposal,
  • "updateVoteThd" - threshold for voting for Cardano SL update proposal.

Please note that values of all thresholds are multiplied by 10⁻¹⁵. So if particular threshold is defined as X, its actual value is X * 10⁻¹⁵.

Section "bootStakeholders" contains bootstrap era stakeholders’ identifiers (StakeholderIds) with corresponding weights.

Field "ftsSeed" contains seed value required for Follow-the-Satoshi mechanism (hex-encoded).

Section "protocolConsts" contains basic protocol constants:

  • "k" - security parameter from the paper,
  • "protocolMagic" - protocol magic section, described fully below:

    • "pm" - protocol magic number,
    • "requiresNetworkMagic" - either "NMMustBeNothing" or "NMMustBeJust",
  • "vssMaxTTL" - VSS certificates maximum timeout to live (number of epochs),
  • "vssMinTTL" - VSS certificates minimum timeout to live (number of epochs).

Section "protocolMagic" defines the protocol magic number. When the protocol magic is changed, all signatures become invalid. This is used to distinguish different networks.

  • "pm" - is the protocol magic number, is included in serialized blocks and headers, and is part of signed data.
  • "requiresNetworkMagic" - will be either "NMMustBeNothing" or "NMMustBeJust"

The "protocolMagic" value can either be an object with the two fields described above, or just a plain integer. In the latter case, "requiresNetworkMagic" will take the default value of "NMMustBeJust".

The "requiresNetworkMagic" setting forms part of the genesis data. However it is configured in the core section.

Section "heavyDelegation" contains an information about heavyweight delegation:

  • "cert" - delegation certificate,
  • "delegatePk" - delegate’s public key,
  • "issuerPk" - stakeholder’s (issuer’s) public key,
  • "omega" - index of epoch the block PSK is announced in, it is needed for replay attack prevention, can be set to arbitrary value in genesis.

Keys in heavyDelegation dictionary are StakeholderIds of certificates’ issuers. They must be consistent with issuerPk values.

Field "startTime" is the timestamp of the 0-th slot. Ideally it should be few seconds later than the cluster actually starts. If it’s significantly later, nodes won’t be doing anything for a while. If it’s slightly before the actual starts, some slots will be missed, but it shouldn’t be critical as long as less than k slots are missed.

Section “vssCerts” contains VSS certificates:

  • "expiryEpoch" - index of epoch until which (inclusive) the certificate is valid;
  • "signature" - signature of certificate,
  • "signingKey" - key used for signing,
  • "vssKey" - VSS public key.

Keys in vssCerts dictionary are StakeholderIds of certificates’ issuers. They must be consistent with signingKey values.

Balances and stakes

In genesis data there are two fields which determine balances: avvmDistr and nonAvvmBalances:

  • The former is a map from AVVM addresses to balances
  • The latter is a map from Cardano-SL addresses to balances

AVVM addresses can be easily converted to Cardano-SL addresses (redeem type), so obtaining a final map from Cardano-SL addresses to balances is trivial.

Stakes are computed from balances w.r.t. bootStakeholders map, because all genesis addresses have BootstraEra stake distribution. Keys in bootStakeholders map are StakeholderIds which will have stake and values are their weights (stake will be distributed proportionally to those weights).

Genesis spec format

Let’s again start with an example:

    genesis:
      spec:
        initializer:
          testBalance:
            poors:        12
            richmen:      4
            richmenShare: 0.99
            useHDAddresses: True
            totalBalance: 600000000000000000
          fakeAvvmBalance:
            count: 10
            oneBalance: 100000
          avvmBalanceFactor: 0.5
          useHeavyDlg: True
          seed: 0
        blockVersionData:
          scriptVersion:     0
          slotDuration:      7000
          maxBlockSize:      2000000
          maxHeaderSize:     2000000
          maxTxSize:         4096 # 4 Kb
          maxProposalSize:   700 # 700 bytes
          mpcThd:            0.01 # 1% of stake
          heavyDelThd:       0.005 # 0.5% of stake
          updateVoteThd:     0.001 # 0.1% of total stake
          updateProposalThd: 0.1 # 10% of total stake
          updateImplicit:    10 # slots
          softforkRule:
            initThd:        0.9 # 90% of total stake
            minThd:         0.6 # 60% of total stake
            thdDecrement:   0.05 # 5% of total stake
          txFeePolicy:
            txSizeLinear:
              a: 155381 # absolute minimal fees per transaction
              b: 43.946 # additional minimal fees per byte of transaction size
          unlockStakeEpoch: 18446744073709551615 # last epoch (maxBound @Word64)
        protocolConstants:
          k: 2
          protocolMagic: 55550001
          vssMinTTL: 2
          vssMaxTTL: 6
        ftsSeed: "c2tvdm9yb2RhIEdndXJkYSBib3JvZGEgcHJvdm9kYSA="
        heavyDelegation: {}
        avvmDistr:
          5ZI8dniHQgyz1N/M6wQa4UDe5imEg/83a1IFmlkP6PI=: 101

Most values are the same as in genesis data. It’s easier to write about differences:

  1. Genesis spec has initializer value not present in genesis data. It’s described in more details below.
  2. Thesholds are specified as floating-point numbers.
  3. ftsSeed is base64-encoded, not hex-encoded.
  4. txFeePolicy is specified slightly differently. The comments in the example explain the format pretty well. Note that the values are lovelaces, not ADA.
  5. Addresses in avvmDistr are presented in base64 format, not base64(url).

Initializer

Sample configuration with initializer was presented above. It contains the following fields:

  • testBalance. It specifies how many nodes should be rich, how many nodes should be poor, how big should be total balance of rich nodes, total balance and whether to use HD addresses or simple PubKey addresses. Assignment of balances and stakes is described in detail below.
  • fakeAvvmBalance. It contains number of fake AVVM seeds and how balance of each seed. These seeds can be used to redeem ADA.
  • avvmBalanceFactor. Real AVVM balances will be multiplied by this factor. Fake AVVM balances are not affected.
  • useHeavyDlg. If it’s True, then for each bootstrap stakeholders there will be a heavyweight delegation certificate from this stakeholder to a core node. Otherwise bootstrap stakeholders will participate in the protocol directly.
  • seed. It is an integer number which will be used to generate all secrets (and consequenlty all public keys, addresses, etc. which are derived from secret keys). It can be overridden using --configuration-seed option.

Initializer makes it quite easy to generate genesis data, because all secrets are derived from only one integer. It’s suitable for custom developer clusters because there we can let some people control all stake. It’s not true for mainnet, where we have three parties, that’s why workflow for mainnet is more complex.

Other features supported by initializer:

  • It allows to have fake AVVM seeds to test ADA redemption, which are needed, because developers and QA specialists don’t know actual seeds having value.
  • It allows to decrease the value of all AVVM seeds using avvmBalanceFactor.
  • It allows to have not only redeem addresses (derived from AVVM seeds), but also ready-to-use public key addresses (specified by testBalance). We can have few addresses with very big balance owned by us and many addresses with small balance shared with everybody (e.g. for faucet).

Balances and stakes

In genesis spec there are two fields which determine balances and stakes: avvmDistr and initializer. avvmDistr is supposed to contain the result of vending (which can be found in lib/configuration.yaml, see mainnet_avvmDistr), but it also can be empty if you don’t need AVVM addresses (or it can be modified in a different way, however you want). Overview of initializer is described above.

In the following, we describe how avvmDistr and nonAvvmBalances (fields in genesis data which determine balances and stakes) are computed from genesis spec.

  • avvmDistr in genesis data depends on avvmDistr from genesis spec and also on fakeAvvmBalance and avvmBalanceFactor from initializer. We generate fakeAvvmBalance.count random seeds, convert them to AVVM addresses and assign each one fakeAvvmBalance.oneBalance coins (lovelaces). avvmDistr is multiplied by avvmBalanceFactor and then is merged with fake distribution.

    Note: if an address is in fake avvm map and in real avvm map, it likely indicates that you are doing something wrong, but the behavior is to pick the fake value.

  • nonAvvmBalances in genesis data depends on testBalance. Let’s describe how balances and addresses generated in this case:

    • To generated balances we first find limit on how much coins we can generate by subtracting sum of all real avvm balances (considering avvmBalanceFactor) from maximal possible coin value (45000000000000000). Total generated value is the minimum between testBalance.totalBalance and the aforementioned limit.

      Let’s say that n is the difference between total generated value and the sum of all fake avvm balances (count * oneBalance). Then we have richmenBalance = richmenShare * n. There will be testBalance.richmen addresses with approximately richmenBalance / testBalance.richmen coins. The rest will be evenly distributed among testBalance.poors addresses.

    • For each richman we generate a simple PubKey addresses without attributes.

      Poor addresses are either HD addresses with empty passphrase or simple PubKey addresses depending on useHDAddresses value. All addresses have BootstrapEra stake distribution.

      Note: rich addresses are derived from secret keys of richmen which may be different from bootstrap stakeholders’ keys. It depends on useHeavyDlg.

System start time

With genesis spec, system start time is taken from command line option (--system-start). The reason why system start is not part of initializer is that dev clusters are to be launched often, and it’s easier to change this value from CLI flags rather then configuration file.

Tools

There are some tools relevant to genesis data.

  • cardano-keygen has a command to dump all genesis secrets to files. Usage: cardano-keygen --system-start 0 --configuration-file <file> --configuration-key <key> --configuration-seed <seed> generate-keys-by-spec --genesis-out-dir <dir>. This command will generate and dump secrets to <dir>. To deploy a cluster you need keys of core nodes. They can be found in <dir>/generated-keys/rich/. It can be used only when genesis spec is used.

  • cardano-keygen also has command to dump genesis data (in JSON format). Usage: cardano-keygen --system-start SYSTEM_START --configuration-file <file> --configuration-key <key> --configuration-seed <seed> dump-genesis-data --path <file>. Default path is genesis-data.json. It can be used to generate genesis data from genesis spec if you want it to be used by nodes or just want to verify generated data.

  • cardano-node-simple and cardano-node also have command line option to dump genesis data. The option is --dump-genesis-data-to genesis.json (data will be dumped to genesis.json).

Generating genesis for testnet

For testnet we can’t use spec for genesis, at least because converting genesis spec to genesis data is currently very slow when there are many HD addresses (and we want to have thousands of HD addresses in testnet). For this reason it’s necessary to generate a JSON file with genesis data first. Fortunately it can be done automatically from genesis spec. The instruction below applies to public testnet, staging testnet, internal staging and other clusters when it’s necessary to generate genesis data from spec.

  1. You need to have a configuration which uses genesis spec. Let’s assume it’s in file CONF_FILE and its key is CONF_KEY. For instance, at the moment of writing, there is internal_staging_gen configuration which can be used to generate genesis for internal_staging_full configuration.
  2. Generate very big random integer. It can be as big as you want. Let’s call it SEED. You can use python, for example: python -c "import secrets; print(secrets.randbelow(2**256))".
  3. Figure out system start time. Ideally it should be set to exactly the time when nodes are launched. In practice it’s impossible. If nodes are deployed before this time, they will just wait. There is nothing bad, except that start is delayed. If nodes are deployed after this time, some slots won’t have blocks. If it’s less than k slots, it shouldn’t be a big problem, blockchain won’t be dead, otherwise you should redeploy with another system start. Let’s call it SYSTEM_START.
  4. Run cardano-keygen --system-start SYSTEM_START --configuration-file CONF_FILE --configuration-key CONF_KEY --configuration-seed CONF_SEED dump-genesis-data --path <file>. Note: this step may take a lot of time (very rough estimation is about 30 seconds for 1000 poor addresses).
  5. After step (4) you have <file> with genesis data. You should put it into repository and change value of file and hash in configuration to this file and its hash (computed using scripts/js/genesis-hash.js). In this example you should update internal_staging_full.core.genesis.src.{file,hash} (updating file is not necessary if you use the same name as in the template).
  6. You also need to know secret keys of core nodes. Use cardano-keygen --system-start 0 --configuration-file CONF_FILE --configuration-seed SEED --configuration-key CONF_KEY generate-keys-by-spec --genesis-out-dir <dir> as described above. Note: you can pass arbitrary system-start here, it doesn’t matter. Note: this step also can take a lot of time.
  7. Save SEED somewhere and make sure it won’t be lost and won’t be known by anybody except people who maintain the testnet.

Generating genesis for mainnet

TODO: we have some scripts and README.md in scripts/prepare-genesis. We probably will never need them anymore, but we need to preserve them to be able to review at any point for example. Probably that README.md should be moved to this section or to this folder (docs/). Some part of that README.md has been moved to this file already.

Parts of configuration file

Core configuration besides genesis

  • requiresNetworkMagic — influences both the genesis data and the address format that the node uses. It can be either:

    • "NMMustBeNothing" (mainnet setting) — means that the protocol magic value will not be included in the address format or transactions.
    • "NMMustBeJust" (public testnet setting, the default) — means that the protocol magic value will be included in the address format and hence transactions.
  • dbSerializeVersionTODO

Infra configuration

TODO

Update configuration

TODO

Ssc configuration

TODO

Delegation configuration

TODO

Block configuration

Block configuration is stored under the <configuration-key>.block key. It contains the following values:

  • networkDiameter (seconds) — how much time it takes to propagate a block across the network. In practice this value determines when slot leaders create a block (it’s done networkDiameter seconds before the end of the slot).
  • recoveryHeadersMessage — how many headers will be sent in a batch. This value should be greater than k.

Node configuration

Node configuration is accessible via the <configuration-key>.node key. It has the following values:

TODO: describe values.

Predefined configurations

This section describes various configuration keys we have in lib/configuration.yaml file. For long-term clusters we also describe differences between that cluster’s configuration and mainnet configuration. Note that generally startTime and protocolMagic are different for all different clusters, so these differences aren’t mentioned. Also update configuration contains versions which are initially set to zeros and then are increased when software/protocol updates are performed. So update configurations also generally differ and this differences isn’t mentioned.

Mainnet configurations

There are several mainnet configurations in lib/configuration.yaml file for different keys.

  • mainnet_base configuration serves as a basis for other mainnet configurations and shouldn’t be used directly. It also can be used as a basis for non-mainnet configurations.
  • mainnet_full configuration is what core nodes use in real mainnet. mainnet_wallet_win64 and mainnet_wallet_macos64 are almost same, but they have different update configurations (different application name, version, etc.). They should be used by wallets (i. e. nodes launched with Daedalus).
  • mainnet_dryrun_full is like mainnet_full, but for staging. mainnet_dryrun_wallet_win64 and mainnet_dryrun_wallet_macos64 are for staging wallets. Staging configuration is very similar to real mainnet configuration, there are few differences:

    • Different bootstrap stakeholders, stake is delegated to different public keys (number of stakeholders is the same).
    • Real mainnet has more fake addresses with 1 ADA (100 vs 6).
  • mainnet_staging_short_epoch_full is needed for testing, because in real mainnet and in staging epoch is very long, but we want to be sure that crossing epoch boundary won’t lead to problems. Cluster with this configuration doesn’t run constantly, it’s launched before preparing installers for release. It is very similar to mainnet_dryrun_full. There are few differences:

    • It has much smaller k which implies much shorter epoch.
    • It also has only 3 core nodes with different keys. Corresponding keys can be found in secrets/mainnet-staging-short-epoch directory.
    • It has more AVVM seeds with ADA for testing.

Using mainnet staging configuration with short epoch

mainnet_staging_short_epoch_full configuration can’t be used as is, because it has hardcoded system start time which should be modified every time we want to launch cluster. In order to use it one needs to:

  1. Figure out system start time.
  2. Modify value of startTime in mainnet-staging-short-epoch-genesis.json.
  3. Use scripts/js/genesis-hash.hs to compute new hash.
  4. Modify value of mainnet_staging_short_epoch_full.core.genesis.src.hash in configuration file to the hash from the previous step.

Internal staging configurations

internal_staging_gen configuration is used to generate genesis data for internal_staging. Please see Generating genesis for testnet section for more details.

internal_staging_full configuration is used by full nodes in internal staging. Currently it’s a stub, because internal staging isn’t deployed yet.

internal_staging_wallet_win64 and internal_staging_wallet_macos64 are like internal_staging_full, but for wallets.

Differences between mainnet and internal_staging:

  • In internal staging all real AVVM balances are multiplied by 0.1.
  • In internal staging there are 100 fake AVVM addresses with 1000000 ADA.
  • In internal staging there are non-avmm balances. A few rich addresses and many poor addresses.

Devnet configuration

There is a configuration called devnet which should be used to setup clusters for developers or QA to test something. There are some values which should be set carefully before using this configuration, they depend on particular task:

  • slotDuration, k are parameters of the protocol, set to reasonable defaults, but sometimes we may want to use different values. Reminder: number of slots in an epoch is 10 · k.
  • recoveryHeadersMessage depends on k. Set their values properly if you modify k. See node configuration.
  • genesis.spec.initializer.richmen is basically the number of core nodes. Should be the same as the number of core nodes deployed.

Other configurations

There are few more configurations which should be briefly mentioned. dev configuration is used by developers to launch cluster locally. It’s very simple and provides bare minimum. bench configuration is used for benchmarks. test configuration is used in tests by default (it’s embedded into binary). default configuration is an alias for dev.