* chore: Set version as v1.27.0-rc1
Set version as v1.27.0-rc1, run make gen & make docsgen-cli
* Update changelog
Update changelog
* Update changelog
Update changelog based on feedback
* nv-skeleton documentation
Add a tutorial for how one can create a nv-skeleton in Lotus
* Add footnote for `Add migration` step
Add footnote for `Add migration` step
* Indent migration-code
Indent migration-code to make it show properly as a footnote.
* Add ref-fvm and filecoin-ffi checklist
Add ref-fvm and filecoin-ffi checklist
* Add Filecoin-FFI steps
Add Filecoin-FFI steps
* Add step to params_butterfly.go
Add step to params_butterfly.go
* Fix typo
Fix typo
* Add links to reference PRs
Add links to reference PRs
* Update ref-fvm list
Update ref-fvm list
The naming of `LookupID` can cause confusion when resolving actor IDs vs
ID addresses. To avoid this:
* Refactor `StateTree` `LookupID` to `LookupIDAddress`, because it
returns ID address.
* Refactor `StateManager` `LookupID` to
`LookupIDAddress` because it also returns ID address via a chain call to
`StateTree`.
* Introduce a new API `StateManager` dedicated to resolving address to
actor ID, called `LookupID` which returns `abi.ActorID`.
For context, see:
* https://github.com/filecoin-project/lotus/pull/11723#discussion_r1534728607
* deps: update dependencies to address migration memory bloat
to address memory concerns during a heavy migration
Ref: https://github.com/filecoin-project/go-state-types/pull/260
Ref: https://github.com/whyrusleeping/cbor-gen/pull/96
Ref: https://github.com/filecoin-project/go-amt-ipld/pull/90
* release: prep v1.26.3 patch
Prep v1.26.3 patch release:
- Update changelog, version and make gen + make docsgen-cli
* deps: update cbor-gen to tagged version
deps: update cbor-gen to tagged version
* deps: update go-state-types to tagged version
deps: update go-state-types to tagged version v0.13.2
* chore: deps: update go-state-types to v0.13.3
Fixes a panic when we have fewer than 1k proposals.
---------
Co-authored-by: Phi-rjan <orjan.roren@gmail.com>
Co-authored-by: Rod Vagg <rod@vagg.org>
Co-authored-by: Steven Allen <steven@stebalien.com>
* deps: update dependencies to address migration memory bloat
to address memory concerns during a heavy migration
Ref: https://github.com/filecoin-project/go-state-types/pull/260
Ref: https://github.com/whyrusleeping/cbor-gen/pull/96
Ref: https://github.com/filecoin-project/go-amt-ipld/pull/90
* release: prep v1.26.3 patch
Prep v1.26.3 patch release:
- Update changelog, version and make gen + make docsgen-cli
* deps: update cbor-gen to tagged version
deps: update cbor-gen to tagged version
* deps: update go-state-types to tagged version
deps: update go-state-types to tagged version v0.13.2
* chore: deps: update go-state-types to v0.13.3
Fixes a panic when we have fewer than 1k proposals.
---------
Co-authored-by: Rod Vagg <rod@vagg.org>
Co-authored-by: Steven Allen <steven@stebalien.com>
* paths: Debugging local storage reservations
* paths: Log when individual reservation is less than on-disk space
* paths: fix debug reservations print
* paths: More reserve logs
* paths: More more reserve logs
* paths: add stacks to duplicate done call log
* curio: task storage: Release storage at most once
* curio: cleanup before restarting sdr
* address review
* paths: Simplify reservation release logic
This is showing up in profiles so I figured I'd optimize it a bit:
1. Avoid holding locks while recording metrics.
2. Slightly reduce allocations by re-using the metrics "mutators".
Also, use the passed context for better tracing.
This is unlikely to make a huge difference, but it may help RPC
providers a _tiny_ bit and doesn't really move the complexity needle.
In https://github.com/filecoin-project/lotus/issues/11703, a user came across an edge case where an SP:
1. Submitted a bad PoSt.
2. Terminated their sectors.
3. Withdrew all funds (after paying all fees).
This left the SP's miner actor with a bad proof but without any funds to pay for disputes. On the other hand, given that the sectors have been terminated and termination fees were already paid we don't really care about fining the SP for the bad proofs anyways.
But... we still don't want to submit dispute messages in this case because doing so isn't free and doesn't really serve a purpose.
So add a check to fetch miner wallet balance and only send disputing messages if balance is non zero.
fixes https://github.com/filecoin-project/lotus/issues/11715
Eth subscribe tipsets API should only return tipsets that have been executed.
We do this by only returning the parent tipset of the latest tipset received by ETH Subscribe from it's TipSetFilter subscription.
Closes#11807
Subsumes #11816
* Unset env-var for TestForkPreMigration
Unset LOTUS_DISABLE_PRE_MIGRATIONS env-var for TestForkPreMigration test, so it does not hang
* Err-check to make lint happy
Err-check to make lint happy
* ci: rename install ubuntu deps to install system dependencies
* ci: migrate the release workflow to github actions
* ci: set permissions required by the newly added workflows explicitly
* ci: prevent duplicate release and docker publishing
* ci: trigger docker workflow on push to master
* ci: do not pass tokens to goreleaser on dry release runs
* ci: specify higher permissions on a job level
* ci: create gh action that installs go from go.mod
* ci: create gh action that installs ubuntu dependencies
* ci: create gh action that starts yugabyte db
* ci: create gh workflow that runs go tests
* ci: test the test workflow
* ci: delete duplicate line from runner configuration
* Revert "ci: test the test workflow"
This reverts commit d1da190c375f7104dc1f5151b6f2e8c405d1fb82.
* ci: match the circleci runners more accurately
* ci: fetch all filecoin-ffi dependencies
* ci: rename fetch/install job to cache to better match underlying tasks
* ci: set permissions required by the test workflow explicitly
* ci: create gh action that installs go from go.mod
* ci: create gh action that installs ubuntu dependencies
* ci: create gh workflow that runs go checks
* ci: test the check workflow
* Revert "ci: test the check workflow"
This reverts commit 1d0759d3c69eae99d17c5a80470b7a40d5b3bc94.
* ci: create gh action that installs go from go.mod
* ci: create gh action that installs ubuntu dependencies
* ci: create gh workflow that runs go build
* ci: test the build workflow
* Revert "ci: test the build workflow"
This reverts commit 8150510aae4d6d1fb1d719e2d69057a3c0695d32.
Codec != Raw likely means built-in actor events, which we can safely skip, so
avoid filling up logs with warnings for every built-in actor event that comes
past this point.
Fixes: https://github.com/filecoin-project/lotus/issues/11718
This is done with the intention to add new {Get,Subscribe}ActorEvents in a
future release (i.e. soon!) with both decoded values (dag-json represented)
and simplified (no flags or codec). But because this comes with some
trade-offs wrt fidelity of information (e.g. likely needing to drop events with
badly encoded values, and not retaining original codec), we need to also have
a Raw form of these APIs for consumers that want to take on the burden of
consuming them as they are.
This enhancement optimizes the schema migration process for the event
index by preventing the redundant execution of Data Definition Language
(DDL) statements that define the event schema. Traditionally, these DDL
statements were grouped into a single slice, reflecting the most current
version of the event index schema. With each migration, this slice was
updated to the latest schema iteration, executing all statements in
bulk. Initially, this method sufficed as migrations were focused on
adding indices to existing table columns.
However, as the database schema evolves to meet new requirements, such
as the forthcoming migrations that involve changes to table schemas
(notably, indexing events by emitter actor ID instead of addresses),
the prior approach of bulk execution of DDL statements becomes
unsuitable: it will no longer be safe to repeatedly execute DDL
statements in previous migrations, because the upcoming one changes
`event` table column structure. To address this issue, the work here has
isolated the event index schema migrations on a per-version basis. This
adjustment ensures that only the necessary DDL statements are executed
during each migration, avoiding the inefficiencies and potential errors
associated with redundant executions.
The work here should also minimize the refactoring required for future
migrations, facilitating a smoother introduction of significant schema
updates.
* [WIP] feat: Add nv22 skeleton
Addition of Network Version 22 skeleton
* update FFI
* feat: drand: refactor round verification
* feat: sealing: Support nv22 DDO features in the sealing pipeline (#11226)
* Initial work supporting DDO pieces in lotus-miner
* sealing: Update pipeline input to operate on UniversalPiece
* sealing: Update pipeline checks/sealing states to operate on UniversalPiece
* sealing: Make pipeline build with UniversalPiece
* move PieceDealInfo out of api
* make gen
* make sealing pipeline unit tests pass
* fix itest ensemble build
* don't panic in SectorsStatus with deals
* stop linter from complaining about checkPieces
* fix sector import tests
* mod tidy
* sealing: Add logic for (pre)committing DDO sectors
* sealing: state-types with method defs
* DDO non-snap pipeline works(?), DDO Itests
* DDO support in snapdeals pipeline
* make gen
* update actor bundles
* update the gst market fix
* fix: chain: use PreCommitSectorsBatch2 when setting up genesis
* some bug fixes
* integration working changes
* update actor bundles
* Make TestOnboardRawPieceSnap pass
* Appease the linter
* Make deadlines test pass with v12 actors
* Update go-state-types, abstract market DealState
* make gen
* mod tidy, lint fixes
* Fix some more tests
* Bump version in master
Bump version in master
* Make gen
Make gen
* fix sender
* fix: lotus-provider: Fix winning PoSt
* fix: sql Scan cannot write to an object
* Actually show miner-addrs in info-log
Actually show miner-addrs in lotus-provider info-log
* [WIP] feat: Add nv22 skeleton
Addition of Network Version 22 skeleton
* update FFI
* ddo is now nv22
* make gen
* temp actor bundle with ddo
* use working go-state-types
* gst with v13 market migration
* update bundle, builtin.MethodsMiner.ProveCommitSectors2 -> 3
* actually working v13 migration, v13 migration itest
* Address review
* sealing: Correct DDO snap pledge math
* itests: Mixed ddo itest
* pipeline: Fix sectorWeight
* sealing: convert market deals into PAMs in mixed sectors
* sealing: make market to ddo conversion work
* fix lint
* update gst
* Update actors and GST to lastest integ branch
* commit batcher: Update ProveCommitSectors3Params builder logic
* make gen
* use builtin-actors master
* ddo: address review
* itests: Add commd assertions to ddo tests
* make gen
* gst with fixed types
* config knobs for RequireActivationSuccess
* storage: Drop obsolete flaky tasts
---------
Co-authored-by: Jennifer Wang <jiayingw703@gmail.com>
Co-authored-by: Aayush <arajasek94@gmail.com>
Co-authored-by: Shrenuj Bansal <shrenuj.bansal@protocol.ai>
Co-authored-by: Phi <orjan.roren@gmail.com>
Co-authored-by: Andrew Jackson (Ajax) <snadrus@gmail.com>
Co-authored-by: TippyFlits <james.bluett@protocol.ai>
* feat: implement FIP-0063
* chore: deps: update to go-multiaddr v0.12.2 (#11602)
* feat: fvm: update the FVM/FFI to v4.1 (#11608) (#11612)
This:
1. Adds nv22 support.
2. Updates the message tracing format.
Co-authored-by: Steven Allen <steven@stebalien.com>
* AggregateProofType nil when doing batch updates
Use latest nv22 go-state-types version with matching update
* Update to v13.0.0-rc.2 bundle
* chore: Upgrade heights and codename
Update upgrade heights
Co-Authored-By: Steven Allen <steven@stebalien.com>
* Update epoch after nv22 DRAND switch
Update epoch after nv22 DRAND switch
* Update Mango codename to Phoneix
Make the codename for the Drand-change inline with Dragon style.
* Add UpgradePhoenixHeight to API params
* set UpgradePhoenixHeight to be one hour after Dragon
* Make gen
Make gen and UpgradePhoenixHeight in butterfly and local devnet to be in line with Calibration and Mainnet
* Update epoch heights (#11637)
Update epoch heights
* new: add forest bootstrap nodes (#11636)
Signed-off-by: samuelarogbonlo <sbayo971@gmail.com>
* Merge pull request #11491 from filecoin-project/fix/remove-decommissioned-pl-bootstrap-nodes
Remove PL operated bootstrap nodes from mainnet.pi
* feat: api: new verified registry methods to get all allocations and claims (#11631)
* new verireg methods
* update changelog and add itest
* update itest and cli
* update new method's support till v9
* remove gateway APIs
* fix cli internal var names
* chore:: backport #11609 to the feat/nv22 branch (#11644)
* feat: api: improve the correctness of Eth's trace_block (#11609)
* Improve the correctness of Eth's trace_block
- Improve encoding/decoding of parameters and return values:
- Encode "native" parameters and return values with Solidity ABI.
- Correctly decode parameters to "create" calls.
- Use the correct (ish) output for "create" calls.
- Handle all forms of "create".
- Make robust with respect to reverts:
- Use the actor ID/address from the trace instead of looking it up in
the state-tree (may not exist in the state-tree due to a revert).
- Gracefully handle failed actor/contract creation.
- Improve performance:
- We avoid looking anything up in the state-tree when translating the
trace, which should significantly improve performance.
- Improve code readability:
- Remove all "backtracking" logic.
- Use an "environment" struct to store temporary state instead of
attaching it to the trace.
- Fix random bugs:
- Fix an allocation bug in the "address" logic (need to set the
capacity before modifying the slice).
- Improved error checking/handling.
- Use correct types for `trace_block` action/results (create, call, etc.).
- And use the correct types for Result/Action structs instead of reusing the same "Call" action every time.
- Improve error messages.
* Make gen
Make gen
---------
Co-authored-by: Steven Allen <steven@stebalien.com>
* fix: add UpgradePhoenixHeight to StateGetNetworkParams (#11648)
* chore: deps: update to go-state-types v13.0.0-rc.1
* do NOT update the cache when running the real migration
* Merge pull request #11632 from hanabi1224/hm/drand-test
feat: drand quicknet: allow scheduling drand quicknet upgrade before nv22 on 2k devnet
* chore: deps: update to go-state-types v13.0.0-rc.2
chore: deps: update to go-state-types v13.0.0-rc.2
* feat: set migration config UpgradeEpoch for v13 actors upgrade
* Built-in actor events first draft
* itest for DDO non-market verified data w/ builtin actor events
* Tests for builtin actor events API
* Clean up DDO+Events tests, add lots of explainer comments
* Minor tweaks to events types
* Avoid duplicate messages when looking for receipts
* Rename internal events modules for clarity
* Adjust actor event API after review
* s/ActorEvents/Events/g in global config
* Manage event sending rate for SubscribeActorEvents
* Terminate SubscribeActorEvents chan when at max height
* Document future API changes
* More clarity in actor event API docs
* More post-review changes, lots of tests for SubscribeActorEvents
Use BlockDelay as the window for receiving events on the SubscribeActorEvents
channel. We expect the user to have received the initial batch of historical
events (if any) in one block's time. For real-time events we expect them to
not fall behind by roughly one block's time.
* Remove duplicate code from actor event type marshalling tests
Reduce verbosity and remove duplicate test logic from actor event types
JSON marshalling tests.
* Rename actor events test to follow go convention
Add missing `s` to `actor_events` test file to follow golang convention
used across the repo.
* Run actor events table tests in deterministic order
Refactor `map` usage for actor event table tests to ensure deterministic
test execution order, making debugging potential issues easier. If
non-determinism is a target, leverage Go's built-in parallel testing
capabilities.
* Reduce scope for filter removal failure when getting actor events
Use a fresh context to remove the temporary filter installed solely to
get the actor events. This should reduce chances of failure in a case
where the original context may be expired/cancelled.
Refactor removal into a `defer` statement for a more readable, concise
return statement.
* Use fixed RNG seed for actor event tests
Improve determinism in actor event tests by using a fixed RNG seed. This
makes up a more reproducible test suit.
* Use provided libraries to assert eventual conditions
Use the functionalities already provided by `testify` to assert eventual
conditions, and remove the use of `time.Sleep`.
Remove duplicate code in utility functions that are already defined.
Refactor assertion helper functions to use consistent terminology:
"require" implies fatal error, whereas "assert" implies error where the
test may proceed executing.
* Update changelog for actor events APIs
* Fix concerns and docs identified by review
* Update actor bundle to v13.0.0-rc3
Update actor bundle to v13.0.0-rc3
* Prep Lotus v1.26.0-rc1
- For sanity reverting the mainnet upgrade epoch to 99999999, and then only set it when cutting the final release
-Update Calibnet CIDs to v13.0.0-rc3
- Add GetActorEvents, SubscribeActorEvents, GetAllClaims and GetAllAllocations methods to the changelog
Co-Authored-By: Jiaying Wang <42981373+jennijuju@users.noreply.github.com>
* Update CHANGELOG.md
Co-authored-by: Masih H. Derkani <m@derkani.org>
* Make gen
Make gen
* fix: beacon: validate drand change at nv16 correctly
* bump to v1.26.0-rc2
* test: cleanup ddo verified itest, extract steps to functions
also add allocation-removed event case
* test: extract verified DDO test to separate file, add more checks
* test: add additional actor events checks
* Add verification for "deal-activated" actor event
* docs(drand): document the meaning of "IsChained" (#11692)
* Resolve conflicts
I encountered multiple issues when trying to run make gen. And these changes fixed a couple of them:
- go mod tidy
- Remove RaftState/RaftLeader
- Revert `if ts.Height() > claim.TermMax+claim.TermStart || !cctx.IsSet("expired")` to the what is in the release/v1.26.0: `if tsHeight > val.TermMax || !expired`
* fixup imports, make jen
* Update version
Update version in master to v1.27.0-dev
* Update node/impl/full/dummy.go
Co-authored-by: Łukasz Magiera <magik6k@users.noreply.github.com>
* Adjust ListClaimsCmd
Adjust ListClaimsCmd according to review
---------
Signed-off-by: samuelarogbonlo <sbayo971@gmail.com>
Co-authored-by: TippyFlits <james.bluett@protocol.ai>
Co-authored-by: Aayush <arajasek94@gmail.com>
Co-authored-by: Łukasz Magiera <magik6k@users.noreply.github.com>
Co-authored-by: Jennifer Wang <jiayingw703@gmail.com>
Co-authored-by: Shrenuj Bansal <shrenuj.bansal@protocol.ai>
Co-authored-by: Andrew Jackson (Ajax) <snadrus@gmail.com>
Co-authored-by: Steven Allen <steven@stebalien.com>
Co-authored-by: Rod Vagg <rod@vagg.org>
Co-authored-by: Samuel Arogbonlo <47984109+samuelarogbonlo@users.noreply.github.com>
Co-authored-by: LexLuthr <88259624+LexLuthr@users.noreply.github.com>
Co-authored-by: tom123222 <160735201+tom123222@users.noreply.github.com>
Co-authored-by: Aarsh Shah <aarshkshah1992@gmail.com>
Co-authored-by: Masih H. Derkani <m@derkani.org>
Co-authored-by: Jiaying Wang <42981373+jennijuju@users.noreply.github.com>
Add tests that assert the embedded built-in actors metadata is correct:
* the corresponding CAR file is present in built-in actors released
assets as a CAR file.
* manifest CID is the only root CID in the corresponding CAR file.
* actor CIDs are present in the corresponding CAR file.
Fixes#11683
These are the folks who are actively contributing to lotus maintenance, issue triaging, PR reviews and development.
Made this change to decouple lotus ownership with other filecoin core repo's ownership and maintainership that comes with lotus-maintainer perm
Now that 1.22 is out. Libp2p will also be dropping support for 1.20
soon (if it hasn't already?) so it can _finally_ stop supporting
specific compiler versions (we can stop caring so much about the
"maximum" supported go version).
* Improve the correctness of Eth's trace_block
- Improve encoding/decoding of parameters and return values:
- Encode "native" parameters and return values with Solidity ABI.
- Correctly decode parameters to "create" calls.
- Use the correct (ish) output for "create" calls.
- Handle all forms of "create".
- Make robust with respect to reverts:
- Use the actor ID/address from the trace instead of looking it up in
the state-tree (may not exist in the state-tree due to a revert).
- Gracefully handle failed actor/contract creation.
- Improve performance:
- We avoid looking anything up in the state-tree when translating the
trace, which should significantly improve performance.
- Improve code readability:
- Remove all "backtracking" logic.
- Use an "environment" struct to store temporary state instead of
attaching it to the trace.
- Fix random bugs:
- Fix an allocation bug in the "address" logic (need to set the
capacity before modifying the slice).
- Improved error checking/handling.
- Use correct types for `trace_block` action/results (create, call, etc.).
- And use the correct types for Result/Action structs instead of reusing the same "Call" action every time.
- Improve error messages.
We did this to avoid using cached migrations in the borked watermelon
upgrade on calibrationnet, but we don't need to keep this in the
code (even for users syncing from scratch).
the .gz files are impossible to rebase/merge/etc. but the files are needed in
the repo for `go build` because they are embeds; so we just don't gzip them
This will re-execute tipsets to forcibly re-compute and store events
when they're missing. This is effectively lazy backfilling of events.
NOTE: This _won't_ backfill the index itself, it'll just give us the
events.
fixes#11335
* chore: cleanup sync serve and reduce log noise
1. Demote a noisy blocksync request error to debug. All this warning
means is that someone is requesting a tipset we don't have.
2. Add a separate warning if we fail to collect a chain. If we have the
tipsets but fail to collect the chain, something is actually wrong.
3. Fix a TODO and return a single CompactedMessages rather than 4
separate values.
* generally reduce the warning to info
It turns out we do fail to gather messages frequently as well, likely
because we have written the tipsets but haven't fetched the messages...
We can now atomically switch chains when checkpointing. Previously,
we'd call `SetHead` followed by `SetCheckpoint`. Unfortunately, that's
not atomic and the "head" could have reverted before we called
`SetCheckpoint` (causing the latter to fail).
Now, we just call `SetCheckpoint` and let `SetCheckpoint` adjust our
head. This changes the behavior of `ChainStore.SetCheckpoint`, but
`Syncer.SyncCheckpoint` is the only caller anyways.
1. Let the user specify `--remove-existing-chain=false` to avoid the
prompt and fail if there's an existing chain.
2. Only prompt when interactive. Otherwise, _don't_ remove the chain
by-default.
The correct name for this field is 'input' according to the Ethereum specs [0].
However, for the longest time, clients have been using 'data' and servers have been
lenient to accept both, preferring 'input' over 'data' when both appear.
Our lack of support for 'input' had gone unnoticed until go-ethereum decided
to adjust their ethclient implementation to issue eth_call and eth_estimateGas
requests with the 'input' field instead of 'data' [1]. This suddenly broke apps
using this client against Lotus' Eth API.
[0]: https://github.com/ethereum/execution-apis/blob/main/src/schemas/transaction.yaml#L33-L35
[1]: ethereum/go-ethereum#28078
Co-authored-by: raulk <raul.kripalani@gmail.com>
Previously, we weren't checking the "to" address. I've also re-ordered
the operations in this function to make it easier to reason about them.
It'll have a slight runtime cost (we _always_ ABI-encode the parameters,
then throw away the result if it turns out we're actually dealing with
an Ethereum transaction), but it's _much_ simpler.
Also explicitly limit how many bytes we're willing to read in one go
such that we're capable of reading a worst-case tipset (like, really,
never going to happen worst-case). Previously, this wasn't an issue.
However, we've bumped the max number of messages from 8,192 to 150,000
and need to limit allocations somewhere else.
