[DO NOT SQUASH MERGE] merge main@35c7c26 to arb0 #16679
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cp from 3.1 --------- Co-authored-by: JkLondon <[email protected]>
Co-authored-by: JkLondon <[email protected]> Co-authored-by: JkLondon <[email protected]>
Follow up #16118. The outer `evm` wasn't used, so it was pointless to cancel it.
introduces `Spec` type gathering all chain specifications (config, genesis, hashes, name, dns, bootnodes) into one type accessible directly and indirectly by chainname. Improves framework of adding new chain specs for Polygon (and arbitrum) --------- Co-authored-by: alex <[email protected]>
Based on this draft PR : #15303 --------- Co-authored-by: Mark Holt <[email protected]> Co-authored-by: antonis19 <[email protected]>
Reverts #15974 due to testing calls to unregistered chains
…16503) after 26d43d5 `TestMiningBenchmark` started failing consistently during `make test-all-race`, e.g. run: https://github.com/erigontech/erigon/actions/runs/16806309189/job/47599533501 upon inspection, it looked like the test was taking 2 mins to re-try mining loop - check `Start mining based on miner.recommit` below: <img width="1088" height="140" alt="Screenshot 2025-08-07 at 18 55 22" src="https://github.com/user-attachments/assets/b3a1cc47-4b4e-47c1-8385-7e2dbeefc173" /> this PR fixes this by setting the miner recommit to 3 secs (our default) also it sets some ports to :0 to let the OS handle these and avoid using fixed port nums in CI
basically the API was returning null for unset blob schedule and lh was shitting the bed
Reverts #16429 just for the time being
References new rpc-tests tag
Should fix #16274 and #16384. Should help with #16151. Also fixes a case where uploading/seeding isn't handled well and caused very high scheduler overhead. It may have caused gradual performance decline over time, I noticed it on long-running nodes. anacrolix/torrent@master...anacrolix There's a few changes of @AskAlexSharov, I will leave them in. Fixes excessive piece hashing routines being spawned. Exposes a few variables that modify webseed HTTP client to help pin down why OVH is being throttled. Fixes a case where torrents added from disk can race against be "required" to proceed past a sync stage. This seemed to cause hashing errors due to races. Fix an annoying log message about `"initialized downloads", "torrents": 0` for snapshots added by non-sync services, that just happen to be adding nothing. Fix a few invariants that weren't checked correctly in anacrolix/torrent. --------- Co-authored-by: alex <[email protected]>
For better troubleshooting it helps to have the full command used + any flags in the logs. This PR prints the startup command while redacting potentially sensitive values like HTTP(S), WS(S) urls and IP addresses. --------- Co-authored-by: antonis19 <[email protected]>
… is done in test-all (#16517) follow up after recent changes which introduced a new workflow for `Test All (with -race)`
Also pulls change in anacrolix/torrent to be less pedantic about chunk size when adding torrents that already exist in the client. Fixes #16481. Needs to be cherry-picked to 3.1 after merge.
First part addressing #16448. #16208 introduced skipping using the Downloader.Completed RPC. This means "completion" flags that the Downloader respects are skipped. Until #15514 those are needed. This introduces the behaviour suggested by #16208 (review). Note that it doesn't *alone* fix #16448. Another change is needed to make sure that torrents with verification requested block Completed too. But there are probably other unexpected behaviours introduced by shortcutting the function as in #16208. After this is main I will cherry pick to 3.1 too (probably combined with the other PR).
``` ## BEFORE Running tool: /usr/bin/go test -benchmem -run=^$ -bench ^BenchmarkPrecompiledEcrecover$ github.com/erigontech/erigon/core/vm -v goos: linux goarch: amd64 pkg: github.com/erigontech/erigon/core/vm cpu: AMD Ryzen 9 7945HX with Radeon Graphics BenchmarkPrecompiledEcrecover BenchmarkPrecompiledEcrecover/CallEcrecoverUnrecoverableKey-Gas=3000 BenchmarkPrecompiledEcrecover/CallEcrecoverUnrecoverableKey-Gas=3000-32 405890 2854 ns/op 3000 gas/op 1051 mgas/s 160 B/op 2 allocs/op BenchmarkPrecompiledEcrecover/ValidKey-Gas=3000 BenchmarkPrecompiledEcrecover/ValidKey-Gas=3000-32 44242 27304 ns/op 3000 gas/op 109.9 mgas/s 224 B/op 4 allocs/op BenchmarkPrecompiledEcrecover/InvalidHighV-bits-1-Gas=3000 BenchmarkPrecompiledEcrecover/InvalidHighV-bits-1-Gas=3000-32 134827530 8.896 ns/op 3000 gas/op 337227 mgas/s 0 B/op 0 allocs/op BenchmarkPrecompiledEcrecover/InvalidHighV-bits-2-Gas=3000 BenchmarkPrecompiledEcrecover/InvalidHighV-bits-2-Gas=3000-32 100000000 11.82 ns/op 3000 gas/op 253796 mgas/s 0 B/op 0 allocs/op BenchmarkPrecompiledEcrecover/InvalidHighV-bits-3-Gas=3000 BenchmarkPrecompiledEcrecover/InvalidHighV-bits-3-Gas=3000-32 96497160 11.92 ns/op 3000 gas/op 251703 mgas/s 0 B/op 0 allocs/op BenchmarkPrecompiledEcrecover/ValidKey2-Gas=3000 BenchmarkPrecompiledEcrecover/ValidKey2-Gas=3000-32 46298 25798 ns/op 3000 gas/op 116.3 mgas/s 224 B/op 4 allocs/op PASS ok github.com/erigontech/erigon/core/vm 9.053s ## AFTER Running tool: /usr/bin/go test -benchmem -run=^$ -bench ^BenchmarkPrecompiledEcrecover$ github.com/erigontech/erigon/core/vm -v goos: linux goarch: amd64 pkg: github.com/erigontech/erigon/core/vm cpu: AMD Ryzen 9 7945HX with Radeon Graphics BenchmarkPrecompiledEcrecover BenchmarkPrecompiledEcrecover/CallEcrecoverUnrecoverableKey-Gas=3000 BenchmarkPrecompiledEcrecover/CallEcrecoverUnrecoverableKey-Gas=3000-32 399147 2857 ns/op 3000 gas/op 1050 mgas/s 160 B/op 2 allocs/op BenchmarkPrecompiledEcrecover/ValidKey-Gas=3000 BenchmarkPrecompiledEcrecover/ValidKey-Gas=3000-32 43558 27176 ns/op 3000 gas/op 110.4 mgas/s 224 B/op 4 allocs/op BenchmarkPrecompiledEcrecover/InvalidHighV-bits-1-Gas=3000 BenchmarkPrecompiledEcrecover/InvalidHighV-bits-1-Gas=3000-32 131973532 9.016 ns/op 3000 gas/op 332745 mgas/s 0 B/op 0 allocs/op BenchmarkPrecompiledEcrecover/InvalidHighV-bits-2-Gas=3000 BenchmarkPrecompiledEcrecover/InvalidHighV-bits-2-Gas=3000-32 120312512 9.970 ns/op 3000 gas/op 300899 mgas/s 0 B/op 0 allocs/op BenchmarkPrecompiledEcrecover/InvalidHighV-bits-3-Gas=3000 BenchmarkPrecompiledEcrecover/InvalidHighV-bits-3-Gas=3000-32 120357236 9.974 ns/op 3000 gas/op 300766 mgas/s 0 B/op 0 allocs/op BenchmarkPrecompiledEcrecover/ValidKey2-Gas=3000 BenchmarkPrecompiledEcrecover/ValidKey2-Gas=3000-32 45979 26090 ns/op 3000 gas/op 115.0 mgas/s 224 B/op 4 allocs/op PASS ok github.com/erigontech/erigon/core/vm 11.094s ``` --------- Co-authored-by: Paweł Bylica <[email protected]>
…options handling (#16505) closes #14283 adds tests in `pool_test.go` for: - pool cleanup - verify both encrypted and decrypted txn pools get cleaned up - blob txns are skipped from shutter pool - if accidentally ending up in the txn pool (matches nethermind) - pool taking into account WithGasTarget and WithTxnIdsFilter options for ProvideTxns fixes for discovered issues: - we weren't updating the cleanup metrics for the encrypted txn pool (made it look like the pool isn't getting cleaned up and leaking mem) - found a bug in `EncryptedTxnSubmissionLess` which is used to traverse the submissions tree - we weren't calling `txnParseCtx.ParseTransaction` with correct input for `hasEnvelope` - we weren't taking into account gas target and txn ids filter options - we had incorrect logic for signing blob txns (there were no usages for it across the codebase before these new tests were added) - added a error log in case the pool unexpectedly stopped due to some err and ProvideTxns was called - makes it much easier to see that the background goroutines are no longer operational - simplified logic for MaxDecryptionKeysDelay (better protection for very long waits for unexpected future slot inputs) - fixes an issue that caused our validators to crash a month ago: ``` [EROR] [07-08|16:59:33.000] background component error err="decryption keys processor issue: decryption keys processing loop: unexpected item txn index lt next txn index: 38275 < 38276" ``` note: im making use of experimental (go1.24) [synctest](https://go.dev/blog/synctest) - it will be added as stable in go1.25
I had started a clarification of Alex's numbered upgrade process. Not sure where the other stuff I replaced got in. Also add a note on torrent logging and how to reset snapshots. Co-authored-by: Alex Sharov <[email protected]>
Co-authored-by: antonis19 <[email protected]>
Implements logic from #16016 to main branch
moved `salt` out of config objects and `dirs` too (not compile-time-known field)
genesis write and state root calc: - it's high-level rare-used logic: mostly `backend.go` calling it - but it's very complex (high-level): creating temporary db, calculating state root, etc... - moving it out of `core` package. because `core` package is low-level - also I moved `ReadChainConfig`/`WriteChainConfig` from `core` to `rawdb` package (In next PR I planing add blockFiles to `temporal.DB` - so, it will manage open/close transactions on this files too - same as currently it does for state files)
move more things to `statecfg` package: - Schema - DomainCfg/IICfg - RegisterDomain calls - AddDependencyBtwnDomains calls - versions about `statecfg` package: all our components have separated cfg package - `txpoolcfg`, `downloadercfg`, `rpccfg`. Because configs: are low-level concept and passing everywhere around APP. Usually such `cfg` packages don't depend on anything big - and have basic types config-types, Default values for configs, constants, etc... Usually `cfg` are known at comp-time (or at least at `init`). `db/kv` aslo as `cfg`-style package - it has configs (`tables.go`) and interfaces.
…ebug` to `dbg` package (#16849)
fixes #16852 checking the `blackListForPruning` was accidentally removed in #16648 diff of Downloader.Add logs (before/after fix): https://www.diffchecker.com/MoE9EsyG/
Co-authored-by: JkLondon <[email protected]>
Co-authored-by: JkLondon <[email protected]>
Co-authored-by: JkLondon <[email protected]>
Co-authored-by: JkLondon <[email protected]>
Main rebase of #16870.
…16863) Co-authored-by: JkLondon <[email protected]>
@yperbasis what do you think? it's `discv4` observer of p2p network to gather statistic and topology. no updates since 2023 and no unit-tests.
Change the channel to a buffered channel with capacity 1. This ensures that the goroutine can always complete the send, even if the caller has already returned due to context cancellation. The response will simply sit in the channel until it is garbage collected, avoiding the leak.
…16823) Switch boundary check from txNum > max to txNum < _min to correctly detect transitions between blocks during descending iteration over history index. This ensures per-block Min/Max are recomputed when crossing into an older block, making the “no system txs” check reliable across the full scan.
yperbasis
changed the title
|
Superseded by PR #17648 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.