github.com/badrootd/nibiru-cometbft@v0.37.5-0.20240307173500-2a75559eee9b/docs/architecture/adr-045-abci-evidence.md (about) 1 # ADR 45 - ABCI Evidence Handling 2 3 ## Changelog 4 * 21-09-2019: Initial draft 5 6 ## Context 7 8 Evidence is a distinct component in a Tendermint block and has it's own reactor 9 for high priority gossipping. Currently, Tendermint supports only a single form of evidence, an explicit 10 equivocation, where a validator signs conflicting blocks at the same 11 height/round. It is detected in real-time in the consensus reactor, and gossiped 12 through the evidence reactor. Evidence can also be submitted through the RPC. 13 14 Currently, Tendermint does not gracefully handle a fork on the main chain. 15 If a fork is detected, the node panics. At this point manual intervention and 16 social consensus are required to reconfigure. We'd like to do something more 17 graceful here, but that's for another day. 18 19 It's possible to fool lite clients without there being a fork on the 20 main chain - so called Fork-Lite. See the 21 [fork accountability](https://github.com/tendermint/tendermint/blob/v0.37.x/spec/light-client/accountability/README.md) 22 document for more details. For a sequential lite client, this can happen via 23 equivocation or amnesia attacks. For a skipping lite client this can also happen 24 via lunatic validator attacks. There must be some way for applications to punish 25 all forms of misbehavior. 26 27 The essential question is whether Tendermint should manage the evidence 28 verification, or whether it should treat evidence more like a transaction (ie. 29 arbitrary bytes) and let the application handle it (including all the signature 30 checking). 31 32 Currently, evidence verification is handled by Tendermint. Once committed, 33 [evidence is passed over 34 ABCI](https://github.com/tendermint/tendermint/blob/v0.37.x/proto/tendermint/abci/types.proto#L354) 35 in BeginBlock in a reduced form that includes only 36 the type of evidence, its height and timestamp, the validator it's from, and the 37 total voting power of the validator set at the height. The app trusts Tendermint 38 to perform the evidence verification, as the ABCI evidence does not contain the 39 signatures and additional data for the app to verify itself. 40 41 Arguments in favor of leaving evidence handling in Tendermint: 42 43 1) Attacks on full nodes must be detectable by full nodes in real time, ie. within the consensus reactor. 44 So at the very least, any evidence involved in something that could fool a full 45 node must be handled natively by Tendermint as there would otherwise be no way 46 for the ABCI app to detect it (ie. we don't send all votes we receive during 47 consensus to the app ... ). 48 49 2) Amensia attacks can not be easily detected - they require an interactive 50 protocol among all the validators to submit justification for their past 51 votes. Our best notion of [how to do this 52 currently](https://github.com/tendermint/tendermint/blob/c67154232ca8be8f5c21dff65d154127adc4f7bb/docs/spec/consensus/fork-detection.md) 53 is via a centralized 54 monitor service that is trusted for liveness to aggregate data from 55 current and past validators, but which produces a proof of misbehavior (ie. 56 via amnesia) that can be verified by anyone, including the blockchain. 57 Validators must submit all the votes they saw for the relevant consensus 58 height to justify their precommits. This is quite specific to the Tendermint 59 protocol and may change if the protocol is upgraded. Hence it would be awkward 60 to co-ordinate this from the app. 61 62 3) Evidence gossipping is similar to tx gossipping, but it should be higher 63 priority. Since the mempool does not support any notion of priority yet, 64 evidence is gossipped through a distinct Evidence reactor. If we just treated 65 evidence like any other transaction, leaving it entirely to the application, 66 Tendermint would have no way to know how to prioritize it, unless/until we 67 significantly upgrade the mempool. Thus we would need to continue to treat evidence 68 distinctly and update the ABCI to either support sending Evidence through 69 CheckTx/DeliverTx, or to introduce new CheckEvidence/DeliverEvidence methods. 70 In either case we'd need to make more changes to ABCI then if Tendermint 71 handled things and we just added support for another evidence type that could be included 72 in BeginBlock. 73 74 4) All ABCI application frameworks will benefit from most of the heavy lifting 75 being handled by Tendermint, rather than each of them needing to re-implement 76 all the evidence verification logic in each language. 77 78 Arguments in favor of moving evidence handling to the application: 79 80 5) Skipping lite clients require us to track the set of all validators that were 81 bonded over some period in case validators that are unbonding but still 82 slashable sign invalid headers to fool lite clients. The Cosmos-SDK 83 staking/slashing modules track this, as it's used for slashing. 84 Tendermint does not currently track this, though it does keep track of the 85 validator set at every height. This leans in favour of managing evidence in 86 the app to avoid redundantly managing the historical validator set data in 87 Tendermint 88 89 6) Applications supporting cross-chain validation will be required to process 90 evidence from other chains. This data will come in the form of a transaction, 91 but it means the app will be required to have all the functionality to process 92 evidence, even if the evidence for its own chain is handled directly by 93 Tendermint. 94 95 7) Evidence from lite clients may be large and constitute some form of DoS 96 vector against full nodes. Putting it in transactions allows it to engage the application's fee 97 mechanism to pay for cost of executions in the event the evidence is false. 98 This means the evidence submitter must be able to afford the fees for the 99 submission, but of course it should be refunded if the evidence is valid. 100 That said, the burden is mostly on full nodes, which don't necessarily benefit 101 from fees. 102 103 104 ## Decision 105 106 The above mostly seems to suggest that evidence detection belongs in Tendermint. 107 (5) does not impose particularly large obligations on Tendermint and (6) just 108 means the app can use Tendermint libraries. That said, (7) is potentially 109 cause for some concern, though it could still attack full nodes that weren't associated with validators 110 (ie. that don't benefit from fees). This could be handled out of band, for instance by 111 full nodes offering the light client service via payment channels or via some 112 other payment service. This can also be mitigated by banning client IPs if they 113 send bad data. Note the burden is on the client to actually send us a lot of 114 data in the first place. 115 116 A separate ADR will describe how Tendermint will handle these new forms of 117 evidence, in terms of how it will engage the monitoring protocol described in 118 the [fork 119 detection](https://github.com/tendermint/tendermint/blob/c67154232ca8be8f5c21dff65d154127adc4f7bb/docs/spec/consensus/fork-detection.md) document, 120 and how it will track past validators and manage DoS issues. 121 122 ## Status 123 124 Proposed. 125 126 ## Consequences 127 128 ### Positive 129 130 - No real changes to ABCI 131 - Tendermint handles evidence for all apps 132 133 ### Neutral 134 135 - Need to be careful about denial of service on the Tendermint RPC 136 137 ### Negative 138 139 - Tendermint duplicates data by tracking all pubkeys that were validators during 140 the unbonding period