You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* Auction resolution latency metric
* Timeboost: swap sequencers seamlessly
* Upgrade actions/upload-artifact in CI
* prioritize reading from timeboostAuctionResolutionTxQueue
* chore: fix some function names in comment
Signed-off-by: linchizhen <[email protected]>
* Add the installation of the cbindgen binary
This binary is used in the make invocation, so it needs to be installed.
* add system test to test seamless swap of active sequencer
* fix lint errors
* Add cbindgen to both workflows
It turns out that the ci.yml workflow was also failing because of the removal of
cbindgen from the Ubuntu 24.04 image used by the GitHub action runners.
This change also moves the installation of cbindgen earlier in the
codequl-analysis.yml file and expands the scope of what is cached after a rust
installation to include the cbindgen binary (which is installed in ~/.cargo/bin)
* Bump golang.org/x/net from 0.26.0 to 0.33.0
Bumps [golang.org/x/net](https://github.com/golang/net) from 0.26.0 to 0.33.0.
- [Commits](golang/net@v0.26.0...v0.33.0)
---
updated-dependencies:
- dependency-name: golang.org/x/net
dependency-type: indirect
...
Signed-off-by: dependabot[bot] <[email protected]>
* Implement CodeAt for contractAdapter
CodeAt is needed if the contract call returns a zero length response.
The bind library code uses CodeAt to check if there is any contract
there at all.
* make redis related updates more async
* Move timeboost init back to mainImpl
It doesn't work where it was before, the contractAdapter is unable to
read the express lane contract.
* address PR comments
* Move express lane start to after init
* Make auctioneer rpc namespace work on non jwt
* make best effort to sync from redis during a swap
* disable control transfer
* update bold
* geth-pin update: performance and metrics improvements
* maintenance api
* reduce roundInfo lock contention in syncFromRedis
* address PR comments
* Flushed TrieDB during maintenance
* Forward auction resolution txs
* Moves maintenance trie cap limit config to execution config
* Fixes units related to trie cap limit config
* Apply suggestions from code review
Co-authored-by: Joshua Colvin <[email protected]>
* Fixes mutex in FlushTrieDB
* maintenance: improve api and address reviews
* address PR comments
* address PR comments
* fix lint error
* fix failing test after merge from upstream
* update pin
* fix for flatcalltracer originally just on v3.3.x branch
fixes NIT-3071
Pulls in OffchainLabs/go-ethereum#401
* comment fixes by PR reviewe
* fix TestPrestateTracingSimple
* further fix TestPrestateTracingSimple
* Add two new fields that were added to the rollup config
* Extra logging for when forwarding fails
* attempt to reduce flakiness of bold virtual block tests
* Option to use redis coordinator to find sequencer
This PR adds two new options for the autonomous-auctioneer which allow
it to discover the active sequencer so that it can send its auction
resolution transactions directly to it. If the sequencer is using the
redis coordinator to manage which sequencer is active, then the same
redis url can be passed as an option to the auctioneer.
The new options are:
--auctioneer-server.redis-coordinator-url
--auctioneer-server.use-redis-coordinator
* revert
* revert changes to nitro-testnode
* allow sequencer to not use block metadata
* do not add zero metadata or track it when timeboost is not enabled
* Use the correct error
When the state provider isn't able to do its job because the chain is lagging
behind head, it's supposed to use a predefined error from the l2stateprovider
package. Instead, the nitro implemention of the provider was using a
redefinition of the error.
While this coding error easily accounts for why there were a bunch of ERROR logs
about the chain's not keeping up with the latest changes, it does not explain
why the test is sometimes flaky. I suspect that the original commits in this PR
were chasing the ERROR logs in the flaky tests, but that those were not the root
cause of the test failures.
* bulk syncing of missing block metadata should start from the block number provided in trackBlockMetadataFrom flag, this prevents querying of stale missing data
* Take out JWT from timeboost system tests
* Add small sleep in BoLD when chain is behind
Pulls in https://github.com/OffchainLabs/bold/pull/726/files
* avoids leaking URL in fallback client
* Fixes import lint issue
* Update bold pin
* TransactionStreamer: dont fetch block metadata when unnecessary
avoid trying to fetch block metadata when nnecessary
* fix failing tests
* fix test
* Revert "Update bold pin"
This reverts commit 0800d46.
* chainInfo supports track-block-metadata-from
* Timeboost: Don't store or publish to feed, blockMetadata of blocks lower than TrackBlockMetadataFrom config option
* allow sequencer to collect metadata without timeboost
* remove accidental nitro-testnode pin update
* add blockinfo to sepolia metadata
* Fix reading of pending ExpressLane messages from redis
* decouple scanning of keys from fetching of messages from redis
* Mark `timeboost` as dangerous since it is a work in progress
Timeboost is not completely finished yet, so not recommended for production use.
* Fix configuration update
* bound the search for pending messages in redis during a switchover
* fix failing tests
* update timeboost config in system_test
* merge conflicts fix
* Do not modify gas cap if already `0` (infinite)
* update pin
* update geth pin
* remove block metadata from sepolia
* Allow BlockMetadata to be generated when timeboost isn't enabled
* fix lint error in systemTest
* address PR comments
* Start tracking metadata in sepolia
Add configuration for block to start tracking sepolia metadata.
* update stake-token to WETH on L1
* Fix syncMonitor's BlockMetadataByNumber response for arb classic block numbers
* Make get_logs call in small chunks
* fix number of block
* Use last confirm assertion in watcher to find the starting point
* update bold
* fix lint
* fix lint
* update default
* Add trackers for missing block metadata retroactively
* Hex encode bidder signatures in S3 published CSVs
Previously the signature field, which is just raw bytes, was being
written directly into the bid history CSVs which are published to S3 by
the auctioneer.
The uploaded files are for consumption by external parties and not used
by the auctioneer after they have been uploaded so no changes are needed
to read the files.
This change is being made pre-mainnet so we don't need to worry about
fixing existing files.
(cherry picked from commit 6a7b125)
* Stylus: visit all relevant contracts if deterministic
* Fix deadlock caused by txStreamer trying to broadcast an executed message during a stopAndWait
* change impl to handle deadlock by draining broadcastChan upon context cancellation
* typo fix
* stop broadcastServer after txStreamer
* document changes
* update buildspec to same as master
* fix dockerfile
* Add new Pectra header
Pull in update to go-ethereum that adds new Pectra header
* [Backport] Fix auction resolution during a tie
* [Backport] Remove timeboost's MaxQueuedTxCount config option
* [Backport] ExpressLane submissions are returned results immediately after they are queued and use channels in redisCoordinator to accept push-updates to redis instead of launching goroutines
* add missing newLines
* [Backport] Implement block-based timeout and set timeboost queued transactions to timeout at the round endtime
* Handle getting creationAtBlock for l3 bold [backport OffchainLabs#2967]
backport OffchainLabs#2967 and update BoLD repo to latest
* Fix backport merge
* Update bold fast confirmation based on legacy fast confirmation improvements
* Memory improvements in DAS REST client
Stream response body to json decoder.
Use cancelable context for http get.
* fix test
* Make config options required for navigating pectra header issues default
* [Backport] Increase default value for timeboost.max-future-sequence-distance to 1000
* Revert "Make config options required for navigating pectra header issues default"
This reverts commit 5439d26.
* [Backport] Express Lane Submission prechecker
Original PR: OffchainLabs#3039
* [Backport] Change timeboost_auctionresolution metric to Gauge
Auction resolution events happen only once per minute so using a
histogram metric for arb_sequencer_timeboost_auction does not really
make sense. Use a gauge intead so we can always see the last resolution
duration.
Originally merged in: OffchainLabs#3067
* Set block to start tracking arb1 block metadata
* [Backport] Rectify evm depth modification while faking a call inside StartTxHook for arb type txs
* Add RequestsHash to gen_header_json.go
* Validate timeboost config options correctly
* address PR comments and remove dangerous before timeboost
* fix ci failure
* Add a node.batch-poster.dangerous.fixed-gas-limit flag
There are rare cases where problems with gas estimation could be malfunctioning
and we want the ability to override it from a config flag. When
--node.batch-poster.dangerous.fixed-gas-limit is non-zero, the value will be
used as a the gas limit when posting batches and will completely bypass gas
estimation.
Resolves: NIT-3225
* batch_poster: override delayed inbox for gat estimation
* batch_poster: update and document gas estimation code
* [Backport] Make some timeboost errors clearer
This is a backport of OffchainLabs#2987
It's not high priority to make it into the release but it makes some
errors seen by express lane controllers a litle clearer.
* add comment
* Fix build
* Comment failing test (in nitro)
* fix ci
* add myself as codeowner
* play with the CI
* Cherry pick network go PR
* Fix go files
* Bring back databases from 2 to 1
---------
Signed-off-by: linchizhen <[email protected]>
Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: Tristan Wilson <[email protected]>
Co-authored-by: Pepper Lebeck-Jobe <[email protected]>
Co-authored-by: Ganesh Vanahalli <[email protected]>
Co-authored-by: Tsahi Zidenberg <[email protected]>
Co-authored-by: linchizhen <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Raul Jordan <[email protected]>
Co-authored-by: Tsahi Zidenberg <[email protected]>
Co-authored-by: Joshua Colvin <[email protected]>
Co-authored-by: Diego Ximenes <[email protected]>
Co-authored-by: Tristan-Wilson <[email protected]>
Co-authored-by: Derek Lee <[email protected]>
Co-authored-by: Aman Sanghi <[email protected]>
Co-authored-by: Pepper Lebeck-Jobe <[email protected]>
Co-authored-by: Sneh Koul <[email protected]>
f.Bool(prefix+".allow-posting-first-batch-when-sequencer-message-count-mismatch", DefaultBatchPosterConfig.Dangerous.AllowPostingFirstBatchWhenSequencerMessageCountMismatch, "allow posting the first batch even if sequence number doesn't match chain (useful after force-inclusion)")
232
+
f.Uint64(prefix+".fixed-gas-limit", DefaultBatchPosterConfig.Dangerous.FixedGasLimit, "use this gas limit for batch posting instead of estimating it")
returnfalse, fmt.Errorf("produced %v blobs for batch but a block can only hold %v (compressed batch was %v bytes long)", len(kzgBlobs), params.MaxBlobGasPerBlock/params.BlobTxBlobGasPerBlob, len(sequencerMsg))
// n.BroadcastServer is stopped after txStreamer and inboxReader because if done before it would lead to a deadlock, as the threads from these two components
1107
+
// attempt to Broadcast i.e send feedMessage to clientManager's broadcastChan when there wont be any reader to read it as n.BroadcastServer would've been stopped
0 commit comments