* Update comments about note commitment tree rebuilds
* Add info-level logs to time note commitment tree rebuilds
* Log the number of rebuilt blocks after a fork, and the time per block
* Move humantime formats to zebra-chain
* Use human-friendly time formatting
* Return the maximum checkpoint height from the chain verifier
* Return the verified block height from the sync downloader
* Track the verified height in the syncer
* Use a lower concurrency limit during full verification
* Get the tip from the state before the first verified block
* Limit the number of submitted download and verify blocks in a batch
* Adjust lookahead limits when transitioning to full verification
* Keep unused extra hashes and submit them to the downloader later
* Remove redundant verified_height and state_tip()
* Split the checkpoint and full verify concurrency configs
* Decrease full verification concurrency to 5 blocks
10 concurrent blocks causes 3 minute stalls on some blocks on my machine.
(And it has about 4x as many cores as a standard machine.)
* cargo +stable fmt --all
* Remove a log that's verbose with smaller lookahead limits
* Apply the full verify concurrency limit to the inbound service
* Add a summary of the config changes to the CHANGELOG
* Increase the default full verify concurrency limit to 30
* Improve time logging using humantime
* Only log full seconds, ignore the fractional part
* Move humantime_seconds to tracing::fmt
* Move the progress task to its own module
* Add missing humantime dependency
* Log the network upgrade in progress logs
* Log when Zebra verifies the final checkpoint
* Decrease the default lookahead limit to 400
* Increase the block verification timeout to 10 minutes
* Halve the default concurrent downloads config
* Try to run the spawned download task before queueing the next download
* Allow verification to be cancelled if the verifier is busy
* Fix the syntax of links in comments
* Fix a mistake in the docs
Co-authored-by: Alfredo Garcia <oxarbitrage@gmail.com>
* Remove unnecessary angle brackets from a link
* Revert the changes for links that serve as references
* Revert "Revert the changes for links that serve as references"
This reverts commit 8b091aa9fab453e7d3559a5d474e0879183b9bfb.
* Remove `<` `>` from links that serve as references
This reverts commit 046ef25620ae1a2140760ae7ea379deecb4b583c.
* Don't use `<` `>` in normal comments
* Don't use `<` `>` for normal comments
* Revert changes for comments starting with `//`
* Fix some warnings produced by `cargo doc`
* Fix some rustdoc warnings
* Fix some warnings
* Refactor some changes
* Fix some rustdoc warnings
* Fix some rustdoc warnings
* Resolve various TODOs
Co-authored-by: teor <teor@riseup.net>
* Fix some unresolved links
* Allow links to private items
* Fix some unresolved links
Co-authored-by: Alfredo Garcia <oxarbitrage@gmail.com>
Co-authored-by: teor <teor@riseup.net>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
* Return BlockDownloadVerifyError from download_and_verify
* Check block requests and genesis for temporary errors
* Ignore DuplicateBlockQueuedForDownload as a temporary error
* Propagate error info to the syncer main loop
* Sleep after temporary genesis download and verify errors
* Increase the block download retry limit to 3
* Remove an obsolete timing check from the tests
We cancel all block downloads for each sync restart,
so we don't need to wait for the maximum hedged download time.
* fix(network): split synthetic NotFoundRegistry from message NotFoundResponse
* docs(network): Improve `notfound` message documentation
* refactor(network): Rename MustUseOneshotSender to MustUseClientResponseSender
```
fastmod MustUseOneshotSender MustUseClientResponseSender zebra*
```
* docs(network): fix a comment typo
* refactor(network): remove generics from MustUseClientResponseSender
* refactor(network): add an inventory collector to Client, but don't use it yet
* feat(network): register missing peer responses as missing inventory
We register this missing inventory based on peer responses,
or connection errors or timeouts.
Inbound message inventory tracking requires peers to send `notfound` messages.
But `zcashd` skips `notfound` for blocks, so we can't rely on peer messages.
This missing inventory tracking works regardless of peer `notfound` messages.
* refactor(network): rename ResponseStatus to InventoryResponse
```sh
fastmod ResponseStatus InventoryResponse zebra*
```
* refactor(network): rename InventoryStatus::inner() to to_inner()
* fix(network): remove a redundant runtime.enter() in a test
* doc(network): the exact time used to filter outbound peers doesn't matter
* fix(network): handle block requests slightly more efficiently
* doc(network): fix a typo
* fmt(network): `cargo fmt` after rename ResponseStatus to InventoryResponse
* doc(test): clarify some test comments
* test(network): test synthetic notfound from connection errors and peer inventory routing
* test(network): improve inbound test diagnostics
* feat(network): add a proptest-impl feature to zebra-network
* feat(network): add a test-only connect_isolated_with_inbound function
* test(network): allow a response on the isolated peer test connection
* test(network): fix failures in test synthetic notfound
* test(network): Simplify SharedPeerError test assertions
* test(network): test synthetic notfound from partially successful requests
* test(network): MissingInventoryCollector ignores local NotFoundRegistry errors
* fix(network): decrease the inventory rotation interval
This stops us waiting 3-4 sync resets (4 minutes) before we retry a missing block.
Now we wait 1-2 sync resets (2 minutes), which is still a reasonable rate limit.
This should speed up syncing near the tip, and on testnet.
* fmt(network): cargo fmt --all
* cleanup(network): remove unnecessary allow(dead_code)
* cleanup(network): stop importing the whole sync module into tests
* doc(network): clarify syncer inventory retry constraint
* doc(network): add a TODO for a fix to ensure API behaviour remains consistent
* doc(network): fix a function doc typo
* doc(network): clarify how we handle peers that don't send `notfound`
* docs(network): clarify a test comment
Co-authored-by: Janito Vaqueiro Ferreira Filho <janito.vff@gmail.com>
Co-authored-by: Janito Vaqueiro Ferreira Filho <janito.vff@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
* refactor(network): rename Advertised to Available
```sh
fastmod Advertised Available zebra*
fastmod advertised available zebra*
```
* refactor(network): allow different available and missing types inside an InventoryStatus
And rename it to ResponseStatus.
Split the methods between ResponseStatus and an InventoryStatus alias.
* refactor(network): add a block_hash convenience method to InventoryHash
* test(network): improve failure logs for connection tests
* fix(inbound): move address sanitization into the response future
* feat(network): send notfound when Zebra doesn't have a block or transaction
* doc(network): move module docs to the top of each module
This makes them more likely to get updated when the module changes.
* fix(network): stop sending unsupported missing inventory types to the registry
* test(network): inbound messages are forwarded to the registry
* test(inbound): test Peers requests to the inbound service, directly and via TCP
* test(network): notfound block responses are sent by the inbound service
* test(network): notfound tx responses are sent by the inbound service
* test(network): increase sync test mock service timeout
The code that these tests use hasn't actually changed much,
and they are only failing on some platforms (coverage, macOS).
So it seems like the extra concurrent inbound tests have pushed them
past their time limit.
(Perhaps due to TCP system calls, or extra serialization work.)
* doc(network): fix typo
Co-authored-by: Janito Vaqueiro Ferreira Filho <janito.vff@gmail.com>
* test(network): remove unnecessary multi-threaded runtime from tests
This prevents `MockService<zebra_state>` timeouts
in the `sync_block_too_high_extend_tips` test,
at the cost of reducing coverage of different execution orders.
Co-authored-by: Janito Vaqueiro Ferreira Filho <janito.vff@gmail.com>
* fix(state): set state concurrency based on other services' concurrency
* fix(sync): increase the sync downloader lookahead limit
It seems like the recent tokio upgrade made this code even more efficient,
so on testnet we can have around 6000 blocks in flight.
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>
* Refactor so that RetryLimit::Future is std::marker::Sync
* Make the syncer future std::marker::Send by spawning tips futures
* Download synced blocks in chain order, not HashSet order
* Improve MockService failure messages
* Add closure-based responses to the MockService API
* Move MockChainTip to zebra-chain
* Add a MockChainTipSender type alias
* Support MockChainTip in ChainSync and its downloader
* Add syncer tests for obtain tips, extend tips, and wrong block hashes
* Add block too high tests for obtain tips and extend tips
* Add syncer tests for duplicate FindBlocks response hashes
* Allow longer request delays for mocked services in syncer tests
* Tweak crawler timings so peers are more likely to be available
* Tweak min peer connection interval so we try all peers
* Let other tasks run between fanouts, so we're more likely to choose different peers
* Let other tasks run between retries, so we're more likely to choose different peers
* Let other tasks run after peer crawler DemandDrop
This makes it more likely that peers will become ready.
* Document the chain verifier
* Drop gossiped blocks that are too far ahead of the tip
* Add extra gossiped block metrics
* Allow extra gossiped blocks, now we have a stricter limit
* Fix a comment
* Check the exact number of blocks in a downloaded block response
* Drop synced blocks that are too far ahead of the tip
* Add extra synced block metrics
* Test dropping gossiped blocks that are too far ahead of the tip
* Allow an extra checkpoint's worth of blocks in the verifier queues
* Actually let's try two extra checkpoints
* Scale extra height limit with lookahead limit
* Also drop blocks that are behind the finalized tip
* Downgrade a noisy log
* Use a debug log for already verified gossiped blocks
* Use debug logs for already verified synced blocks
* Use a named CancelHeartbeatTask unit struct for the channel type
* Prefer cancel handles in selects, if both are ready
* Fix message metrics to just show the command name
* Add metrics for internal requests and responses
* Add internal requests and responses to the messages dashboard
* Add a canceled metric, and peer addresses to request and response metrics
* Add a canceled messages graph
* Add connection state metrics for currently open connections
* Fix the connection state graph with new metrics
* Always send an error before dropping pending responses
* Move error detail logging into `fail_with`
* Delete an unused timer future
* Make error strings in metrics less verbose
* Downgrade some error logs to info
* Remove a redundant expect
* Avoid unnecessary allocations for connection state metrics
* Fix missed updates to mempool and block gossip metrics
* Use a single-thread shared Tokio runtime
This allows it to pause the time and more closely resembles the
environment that's set by default for asynchronous tests.
* Add a `zebra_test::init_async` helper function
Calls `zebra_test::init` but also constructs a single-thread Tokio
runtime and returns it. This makes it simpler to initialize asynchronous
tests that can't use the `#[tokio::test]` attribute.
* Replace usages of `Runtime::new` in tests
Use the new `zebra_test::init_async()` helper function instead.
* Replace `runtime::Builder::new_current_thread()`
Use the new `zebra_test::init_async()` helper function instead.
* Replace `runtime::Builder::new_multi_thread()`
Use the new `zebra_test::init_async()` helper function instead. The test
with the change doesn't necessarily have to use a multi-thread runtime.
* Update `tower` to version `0.4.9`
Update to latest version to add support for Tokio version 1.
* Replace usage of `ServiceExt::ready_and`
It was deprecated in favor of `ServiceExt::ready`.
* Update Tokio dependency to version `1.13.0`
This will break the build because the code isn't ready for the update,
but future commits will fix the issues.
* Replace import of `tokio::stream::StreamExt`
Use `futures::stream::StreamExt` instead, because newer versions of
Tokio don't have the `stream` feature.
* Use `IntervalStream` in `zebra-network`
In newer versions of Tokio `Interval` doesn't implement `Stream`, so the
wrapper types from `tokio-stream` have to be used instead.
* Use `IntervalStream` in `inventory_registry`
In newer versions of Tokio the `Interval` type doesn't implement
`Stream`, so `tokio_stream::wrappers::IntervalStream` has to be used
instead.
* Use `BroadcastStream` in `inventory_registry`
In newer versions of Tokio `broadcast::Receiver` doesn't implement
`Stream`, so `tokio_stream::wrappers::BroadcastStream` instead. This
also requires changing the error type that is used.
* Handle `Semaphore::acquire` error in `tower-batch`
Newer versions of Tokio can return an error if the semaphore is closed.
This shouldn't happen in `tower-batch` because the semaphore is never
closed.
* Handle `Semaphore::acquire` error in `zebrad` test
On newer versions of Tokio `Semaphore::acquire` can return an error if
the semaphore is closed. This shouldn't happen in the test because the
semaphore is never closed.
* Update some `zebra-network` dependencies
Use versions compatible with Tokio version 1.
* Upgrade Hyper to version 0.14
Use a version that supports Tokio version 1.
* Update `metrics` dependency to version 0.17
And also update the `metrics-exporter-prometheus` to version 0.6.1.
These updates are to make sure Tokio 1 is supported.
* Use `f64` as the histogram data type
`u64` isn't supported as the histogram data type in newer versions of
`metrics`.
* Update the initialization of the metrics component
Make it compatible with the new version of `metrics`.
* Simplify build version counter
Remove all constants and use the new `metrics::incement_counter!` macro.
* Change metrics output line to match on
The snapshot string isn't included in the newer version of
`metrics-exporter-prometheus`.
* Update `sentry` to version 0.23.0
Use a version compatible with Tokio version 1.
* Remove usage of `TracingIntegration`
This seems to not be available from `sentry-tracing` anymore, so it
needs to be replaced.
* Add sentry layer to tracing initialization
This seems like the replacement for `TracingIntegration`.
* Remove unnecessary conversion
Suggested by a Clippy lint.
* Update Cargo lock file
Apply all of the updates to dependencies.
* Ban duplicate tokio dependencies
Also ban git sources for tokio dependencies.
* Stop allowing sentry-tracing git repository in `deny.toml`
* Allow remaining duplicates after the tokio upgrade
* Use C: drive for CI build output on Windows
GitHub Actions uses a Windows image with two disk drives, and the
default D: drive is smaller than the C: drive. Zebra currently uses a
lot of space to build, so it has to use the C: drive to avoid CI build
failures because of insufficient space.
Co-authored-by: teor <teor@riseup.net>
* Ignore AlreadyInChain error in the syncer
* Split Cancelled errors; add them to should_restart_sync exceptions
* Also filter 'block is already comitted'; try to detect a wrong downcast
* Implement a task that gossips verified block hashes
* Log an info message for block broadcasts
* Simplify the gossip task
Co-authored-by: Janito Vaqueiro Ferreira Filho <janito.vff@gmail.com>
* Re-use the old tip change if there is no new tip change
Also improve the comments.
* Add an assertion message
* Rename task join handles and futures in start method
* Add a dedicated BlockGossipError type
This type helps distinguish between syncer and state errors.
* Test that committed blocks are gossiped to peers
Also do a minor type cleanup on the existing test code,
replacing `Option<Vec<_>>` with `Vec<_>`.
* Formatting
* Remove excess newlines
Co-authored-by: Alfredo Garcia <oxarbitrage@gmail.com>
* Clear the initial gossiped blocks during test setup
Co-authored-by: Janito Vaqueiro Ferreira Filho <janito.vff@gmail.com>
Co-authored-by: Alfredo Garcia <oxarbitrage@gmail.com>
* Cancel download and verify tasks when the mempool is deactivated
* Refactor enable/disable logic to use a state enum
* Add helper test functions to enable/disable the mempool
* Add documentation about errors on service calls
* Improvements from review
* Improve documentation
* Fix bug in test
* Apply suggestions from code review
Co-authored-by: teor <teor@riseup.net>
Co-authored-by: teor <teor@riseup.net>
* Decide if Zebra is at the chain tip
* Avoid division by zero
* Try increasing EVENT_TIMEOUT
* Increase MAX_TEST_EXECUTION
* Implement basic tests
* Resolve Clippy's erorrs
* change doc comments to normal
Co-authored-by: Alfredo Garcia <oxarbitrage@gmail.com>
* Create a `SyncStatus` helper type
Keeps track if the synchronizer is close to the chain tip or not.
* Refactor `ChainSync` ctor. to return `SyncStatus`
Change the constructor API so that it returns a higher level construct.
* Test if `SyncStatus` waits for the chain tip
Test if waiting for the chain tip to be reached correctly finishes when
the chain tip is reached. This is done by sending recent sync lengths to
the `SyncStatus` instance, and checking that every time a separate
`SyncStatus` instance determines it has reached the tip the original
instance wakes up.
* Add a temporary attribute to allow dead code
The code added isn't used yet, so we'll add a temporary waiver until
another PR is merged to use them.
* Minimal recent sync lengths implementation
Also includes metrics and logging, to make diagnosing bugs easier.
* Add logging to check what happens when Zebra reaches the chain tip
* Add tests for recent sync lengths
- initially empty
- pruned to correct length
- newest entries go first
* Drop a redundant `/` from a Cargo.lock URL
This seems to be a nightly or beta Rust change,
but hopefully stable just accepts it.
* Use metrics histograms to avoid overwriting values
* Add detailed syncer monitoring dashboard
* Increase the recent sync length to 4
This length makes it easier to distinguish between temporary and
sustained errors/syncs.
Co-authored-by: Janito Vaqueiro Ferreira Filho <janito.vff@gmail.com>
* implement and test a rate limit in `request_genesis()`
* add `request_genesis_is_rate_limited` test to sync
* add ensure_timeouts constraint for GENESIS_TIMEOUT_RETRY
* Suppress expected warning logs in zebrad tests
Co-authored-by: teor <teor@riseup.net>
In our README, we tell users to ignore these errors, so we should also
disable the issue URL.
Also include the hash in the error. (We don't want the span active for
all messages, we just want the hash in the error.)
This provides useful and not too noisy output at INFO level. We do an
info-level message on every block commit instead of trying to do one
message every N blocks, because this is useful both for initial block
sync as well as continuous state updates on new blocks.
We should error if we notice that we're attempting to download the same
blocks multiple times, because that indicates that peers reported bad
information to us, or we got confused trying to interpret their
responses.
Using the cancel_handles, we can deduplicate requests. This is
important to do, because otherwise when we insert the second cancel
handle, we'd drop the first one, cancelling an existing task for no
reason.
The hedge middleware implements hedged requests, as described in _The
Tail At Scale_. The idea is that we auto-tune our retry logic according
to the actual network conditions, pre-emptively retrying requests that
exceed some latency percentile. This would hopefully solve the problem
where our timeouts are too long on mainnet and too slow on testnet.
Try to use the better cancellation logic to revert to previous sync
algorithm. As designed, the sync algorithm is supposed to proceed by
downloading state prospectively and handle errors by flushing the
pipeline and starting over. This hasn't worked well, because we didn't
previously cancel tasks properly. Now that we can, try to use something
in the spirit of the original sync algorithm.
This makes two changes relative to the existing download code:
1. It uses a oneshot to attempt to cancel the download task after it
has started;
2. It encapsulates the download creation and cancellation logic into a
Downloads struct.
Conrado Gouvea
teor
Marek
Alfredo Garcia
Janito Vaqueiro Ferreira Filho
Dimitris Apostolou
Henry de Valence