* slasherkv: Set a 1 minute timeout on PruneAttestationOnEpoch operations to prevent very large bolt transactions.
* Fix CI
---------
Co-authored-by: Manu NALEPA <enalepa@offchainlabs.com>
* Migrate Prysm repo to Offchain Labs organization ahead of Pectra upgrade v6
* Replace prysmaticlabs with OffchainLabs on general markdowns
* Update mock
* Gazelle and add mock.go to excluded generated mock file
* adding nil checks on interface
* changelog
* add linting
* adding missed checks
* review feedback
* attestation bits should not be in nil check
* fixing nil checks
* simplifying function
* fixing some missed items
* more missed items
* fixing more tests
* reverting some changes and fixing more tests
* adding in source check back in
* missed test
* sammy's review
* radek feedback
* `PruneProposalsAtEpoch`: Test return value in case of nothing to prune.
* `TestStore_PruneAttestations_OK`: Create unique validator indexes.
Before this commit, `attester2` for `j = n` was the same than
`attester1` for `j = n + 1`, resulting in erasure of a lot of attesters.
I guess it was not the initial intent.
* Slasher pruning: Check if the number of pruned items corresponds to the expectation.
Before this commit, if the pruning function did remove a superset
of the expected pruned items (including all the items), then the test would pass.
* Prune items that should be pruned and stop pruning items that should not be pruned.
The first 8 bytes of the key of `attestation-data-roots` and
`proposal-records` bytes correspond respectively to an encoded epoch and
and encoded slot.
The important word in this sentence is "encoded".
Before this commit, these slot/epoch are SSZ encoded, which means that
they are little-endian encoded.
However:
- `uint64PrefixGreaterThan` uses `bytes.Compare` which expects
big-endian encoded values.
- `for k, _ := c.First(); k != nil; k, _ = c.Next()` iters over the
keys in big-endian order.
The consequence is:
- Some items that should be pruned are not pruned, provoking a disk
usage leak.
- Some items that should not be pruned are pruned, provoking errors like
in https://github.com/prysmaticlabs/prysm/issues/13658.
This commit encodes the slot/epoch as big-endian before storing them in
the database keys.
Why this bug has not been detected in unit test before?
The values used in unit tests before this commit in
`TestStore_PruneProposalsAtEpoch` and `TestStore_PruneAttestations_OK`
are `10` and `20`.
Unfortunately, checking if `littleEndian(20) > littlenEndien(10)`
with the `>` operator considering operands as big-endian encoded returns
the expected result...
Just replacing `20` by `30` trigs the bug.
* Make deepsource happy.
* Slasher: Migrate database from little-endian to big-endian.
* Update beacon-chain/slasher/service.go
Co-authored-by: Preston Van Loon <pvanloon@offchainlabs.com>
* Update beacon-chain/db/slasherkv/migrate.go
Co-authored-by: Preston Van Loon <pvanloon@offchainlabs.com>
* `TestMigrate`: Fix documentation.
---------
Co-authored-by: Preston Van Loon <pvanloon@offchainlabs.com>
* add max/min span visualisation tool cli
* go mod tidy
* lint imports
* remove typo
* fix epoch table value
* fix deepsource
* add dep to bazel
* fix dep import order
* change command name from span to slasher-span-display
* change command args style using - instead of _
* sed s/CONFIGURATION/SLASHER PARAMS//
* change double neg to double pos condition
* remove unused anonymous func
* better function naming
* add range condition
* [deepsource] Fix Empty slice literal used to declare a variable
GO-W1027
* correct typo
* do not show incorrect epochs due to round robin
* fix import
---------
Co-authored-by: Manu NALEPA <enalepa@offchainlabs.com>
* Improve package documentation.
* `processAttestations`: Improve logging.
* Add `Benchmark_checkSurroundVotes` benchmark.
* Implement `saveChunksToDisk` as remplacement of `saveUpdatedChunks`.
The idea is to open only on DB transaction for all validator chunk indexes instead of
one DB transaction per validator chunk index.
It saves the overhead due to transaction start/stop of the DB.
Result of `Benchmark_checkSurroundVotes`:
- Before this commit: 133 seconds
- After this commit: 5.05 seconds
* `LoadSlasherChunks` and `SaveSlasherChunks`: Batch.
* `loadChunks` ==> `loadChunksFromDisk`
* `updatedChunkByChunkIndex`: Don't update if `latestEpochWritten == currentEpoch `.
* `updatedChunkByChunkIndex`: Load all needed chunks once.
* `latestEpochWritten` ==> `latestEpochUpdated`.
* `checkSurroundVotes`: Dump to disk at most every `25_600` chunks.
* `SaveAttestationRecordsForValidators`: Batch.
* `batchSize`: Use as package const and add comment.
* `getChunk` ==> `getChunkFromDatabase`.
* `loadChunks`: Rename variables.
* `Update`: Use explicit arguments.
* `detect_attestations.go`: Reduce abstraction layers.
* `loadAndUpdateChunks`: Change arguments order.
* `updatedChunkByChunkIndex`: Update all known validators in the chunk.
* `LastEpochWrittenForValidators`: Avoid avoidable `for`loop.
* `chunks.go`: Ensure implementations respect the interface.
* `LastEpochWrittenForValidators`: Stop considering lack of epoch as genesis epoch.
* `updatedChunkByChunkIndex`: Don't update latest updated epoch.
And add a bunch of tests.
* Improve slasher cold boot duration.
Before this commit, on a slasher cold boot (aka, without any db),
the `updatedChunkByChunkIndex` function looped for all validators
AND for all epochs between the genesis epoch and the current epoch.
This could take several dozen of minutes, and it is useless since the
min/max spans are actually a circular buffer with a limited lenght.
Cells of min/max spans can be overwritten (with the same value)
plenty of times.
After this commit, the `updatedChunkByChunkIndex` function loops
for all validators AND AT most 'historyLength' lenght.
Every cell of min/max spans are written AT MOST once.
Time needed for slasher boot goes from `O(nm)` to "only" `O(m)`, where:
- `n` is the number of epochs since the genesis.
- `m` is the number of validators.
* First take at updating everything to v5
* Patch gRPC gateway to use prysm v5
Fix patch
* Update go ssz
---------
Co-authored-by: Preston Van Loon <pvanloon@offchainlabs.com>
* `LastEpochWrittenForValidators`: Use golang if.
* `SaveLastEpochsWrittenForValidators`: Refactor.
* `SaveLastEpochsWrittenForValidators`: Fix when `epochByValIndex` > `batchSize`.
Before this commit, `TestStore_LastEpochWrittenForValidators` works if `validatorsCount <= 10000`
and stops working if `validatorsCount > 10000`.
* Slasher: Detect surrounded votes in multiple batches.
Fixes https://github.com/prysmaticlabs/prysm/issues/13591.
* `Test_processAttestations`: Remove duplicated tests.
* Sort indexed attestations by data root.
* `processAttestations`: Don't return duplicate slashings anymore.
Fix https://github.com/prysmaticlabs/prysm/issues/13592.
* `AttesterDoubleVote`: Rename fields.
* Detect double votes in different batches.
In order to do that:
1. Each attestation of the batch is tested against the other attestations of the batch.
2. Each attestation of the batch is tested against the content of the database.
2. Attestations are saved into the database.
Fixes https://github.com/prysmaticlabs/prysm/issues/13590.
* `helpers.go`: Improve naming consistency.
* `detect_attestations.go`: Improve readability.
* `receive.go`: Add `attsQueueSize` in log message.
* `checkSlashableAttestations`: Improve logging.
`avgBatchProcessingTime` is not displayed any more if not batch is
processed.
* `loadChunks`: Use explicit `chunkKind` and `chunkIndices`.
* `getChunk`: Use specific `chunkIndex` and `chunkKind`.
* `validatorIndicesInChunk` -> `validatorIndexesInChunk`.
* `epochUpdateForValidator`: Use explicit arguments.
* `getChunk`: Change order of arguments.
* `latestEpochWrittenForValidator`: Use `ok` parameter.
So the default value is not any more considered as the absence of
value.
* `applyAttestationForValidator`: Use explicit arguments.
* `updateSpans`: Use explicit arguments.
* `saveUpdatedChunks`: Use explicit arguments.
* `checkSurrounds`: Use explicit arguments.
We see here that, previously, in `checkSlashableAttestations`,
`checkSurrounds` was called with the default value of `slashertypes`: `MinSpan`.
Now, we set it expliciterly at `MinSpan`, which may explicit a bug.
* `epochUpdateForValidator`: Set modified by the function argument first.
* `applyAttestationForValidator`: Set mutated argument `chunksByChunkIdx`first.
* `applyAttestationForValidator`: Rename variables.
* `Test_processQueuedAttestations`: Fix test.
Two tests were actually exactly the same.
* `updateSpans`: Keep happy path in the outer scope.
Even if in this case the "happy" path means slashing.
* `checkSurrounds`: Rename variable.
* `getChunk`: Avoid side effects.
It adds a few lines for callers, but it does not modify any more
arguments and it does what it says: getting a chunk.
* `CheckSlashable`: Flatten.
* `detect_attestations_test.go`: Simplify.
* `CheckSlashable`: Add error log in case of missing attestation.
* `processQueuedAttestations`: Extract a sub function.
So testing will be easier.
* `processAttesterSlashings` and `processProposerSlashings`: Improve.
* `processAttesterSlashings`: Return processed slashings.
* `createAttestationWrapper`: Rename variables.
* `signingRoot` ==> `headerRoot` or `dataRoot`.
Before this commit, there is two typse of `signing root`s floating around.
- The first one is a real signing root, aka a hash tree root computed from an object root and
a domain. This real signing root is the object ready to be signed.
- The second one is a "false" signing root, which is actually just the hash tree root of an object. This object is either the `Data` field of an attestation, or the `Header` field of a block.
Having 2 differents objects with the same name `signing root` is quite confusing.
This commit renames wrongly named `signing root` objects.
* `createAttestationWrapper` => `createAttestationWrapperEmptySig`.
So it's clear for the user that the created attestation wrapper has an empty signature.
* Implement `createAttestationWrapper`.
* `processAttestations`: Return processed attester slashings.
* Test `processAttestations` instead of `processQueuedAttestations`.
By testing `processAttestations` instead of `processQueuedAttestations`, we get rid of a lot of tests fixtures, including the 200 ms sleep.
The whole testing duration is shorter.
* `Test_processAttestations`: Allow multiple steps.
* `Test_processAttestations`: Add double steps tests.
Some new failing tests are commented with a corresponding github issue.
* `NextChunkStartEpoch`: Fix function comment.
Co-authored-by: Preston Van Loon <pvanloon@offchainlabs.com>
* `chunks.go`: Avoid templating log messages.
* `checkSlashableAttestations`: Simplify duration computation.
---------
Co-authored-by: Preston Van Loon <pvanloon@offchainlabs.com>
* Slasher: Ensure all gorouting are stopped before running `Stop` actions.
Fixes#13550.
In tests, `exitChan` are now useless since waitgroup are used to wait
for all goroutines to be stopped.
* `slasher.go`: Add comments and rename some variables. - NFC
* `detect_blocks.go`: Improve. - NFC
- Rename some variables.
- Add comments.
- Use second element of `range` when possible.
* `chunks.go`: Remove `_`receivers. - NFC
* `validateAttestationIntegrity`: Improve documentation. - NFC
* `filterAttestations`: Avoid `else`and rename variable. - NFC
* `slasher.go`: Fix and add comments.
* `SaveAttestationRecordsForValidators`: Remove unused code.
* `LastEpochWrittenForValidators`: Name variables consistently. - NFC
Avoid mixes between `indice(s)`and `index(es)`.
* `SaveLastEpochsWrittenForValidators`: Name variables consistently. - NFC
* `CheckAttesterDoubleVotes`: Rename variables and add comments. - NFC
* `schema.go`: Add comments. - NFC
* `processQueuedAttestations`: Add comments. - NFC
* `checkDoubleVotes`: Rename variable. - NFC
* `Test_processQueuedAttestations`: Ensure there is no error log.
* `shouldNotBeSlashable` => `shouldBeSlashable`
* `Test_processQueuedAttestations`: Add 2 test cases:
- Same target with different signing roots
- Same target with same signing roots
* `checkDoubleVotesOnDisk` ==> `checkDoubleVotes`.
Before this commit, `checkDoubleVotes` did two tasks:
- Checking if there are any slashable double votes in the input
list of attestations with respect to each other.
- Checking if there are any slashable double votes in the input
list of attestations with respect to our database.
However, `checkDoubleVotes` is called only in
`checkSlashableAttestations`.
And `checkSlashableAttestations` is called only in:
- `processQueuedAttestations`, and in
- `IsSlashableAttestation`
Study of case `processQueuedAttestations`:
---------------------------------------------
In `processQueuedAttestations`, `checkSlashableAttestations`
is ALWAYS called after
`Database.SaveAttestationRecordsForValidators`.
It means that, when calling `checkSlashableAttestations`,
`validAtts` are ALREADY stored in the DB.
Each attestation of `validAtts` will be checked twice:
- Against the other attestations of `validAtts` (the portion of
deleted code)
- Against the content of the database.
One of those two checks is redundent.
==> We can remove the check against other attestations in `validAtts`.
Study of case `Database.SaveAttestationRecordsForValidators`:
----------------------------------------------------------------
In `Database.SaveAttestationRecordsForValidators`,
`checkSlashableAttestations` is ALWAYS called with a list of
attestations containing only ONE attestation.
This only attestaion will be checked twice:
- Against itself, and an attestation cannot conflict with itself.
- Against the content of the database.
==> We can remove the check against other attestations in `validAtts`.
=========================
In both cases, we showed that we can remove the check of attestation
against the content of `validAtts`, and the corresponding test
`Test_checkDoubleVotes_SlashableInputAttestations`.
* `Test_processQueuedBlocks_DetectsDoubleProposals`: Wrap proposals.
So we can add new proposals later.
* Fix slasher multiple proposals false negative.
If a first batch of blocks is sent with:
- validator 1 - slot 4 - signing root 1
- validator 1 - slot 5 - signing root 1
Then, if a second batch of blocks is sent with:
- validator 1 - slot 4 - signing root 2
Because we have two blocks proposed by the same validator (1) and for
the same slot (4), but with two different signing roots (1 and 2), the
validator 1 should be slashed.
This is not the case before this commit.
A new test case has been added as well to check this.
Fixes#13551
* `params.go`: Change comments. - NFC
* `CheckSlashable`: Keep the happy path without indentation.
* `detectAllAttesterSlashings` => `checkSurrounds`.
* Update beacon-chain/db/slasherkv/slasher.go
Co-authored-by: Sammy Rosso <15244892+saolyn@users.noreply.github.com>
* Update beacon-chain/db/slasherkv/slasher.go
Co-authored-by: Sammy Rosso <15244892+saolyn@users.noreply.github.com>
* `CheckAttesterDoubleVotes`: Keep happy path without indentation.
Well, even if, in our case, "happy path" mean slashing.
* 'SaveAttestationRecordsForValidators': Save the first attestation.
In case of multiple votes, arbitrarily save the first attestation.
Saving the first one in particular has no functional impact,
since in any case all attestations will be tested against
the content of the database. So all but the first one will be
detected as slashable.
However, saving the first one and not an other one let us not
to modify the end to end tests, since they expect the first one
to be saved in the database.
* Rename `min` => `minimum`.
Not to conflict with the new `min` built-in function.
* `couldNotSaveSlashableAtt` ==> `couldNotCheckSlashableAtt`
---------
Co-authored-by: Sammy Rosso <15244892+saolyn@users.noreply.github.com>
* replace eth2 types
* replace protos
* regen proto
* replace
* gaz
* deps
* amend
* regen proto
* mod
* gaz
* gaz
* ensure build
* ssz
* add dep
* no more eth2 types
* no more eth2
* remg
* all builds
* buidl
* tidy
* clean
* fmt
* val serv
* gaz
Co-authored-by: Preston Van Loon <preston@prysmaticlabs.com>
* Value assigned to a variable is never read before being overwritten
* The result of append is not used anywhere
* Suspicious assignment of range-loop vars detected
* Unused method receiver detected
* Revert "Auxiliary commit to revert individual files from 54edcb445484a2e5d79612e19af8e949b8861253"
This reverts commit bbd1e1beabf7b0c5cfc4f514dcc820062ad6c063.
* Method modifies receiver
* Fix test
* Duplicate imports detected
* Incorrectly formatted error string
* Types of function parameters can be combined
* One more "Unused method receiver detected"
* Unused parameter detected in function
* first
* add receive details
* ensure most builds
* add slasherkv changes
* db iface additions
* build
* gaz
* proper todo comment
* terence comments
* sig check
* bad sig checks
* proper lock issue
* fix test
* fix up tests
