The behavior of a request for a UTXO from a previous block depends on
whether that block has already been submitted to the state, or not:
* if it has, the state should be able to find it and answer immediately.
* if it has not, the state should see it in a later request.
However, the previous code only checked committed blocks, not queued
blocks, so if the block containing the UTXO had already arrived but had
not been committed, it would never be scanned.
This patch fixes the problem but is a bad solution, duplicating
computation between the block verifier and the state. A better fix
follows in the next commit.
Make tracing messages more concise by omitting information already
contained in a parent span and by shortening messages. This makes them
easier to read.
Previously, this function was instrumented with a span containing the
parent hash that was the entry to the function. But it doesn't make
sense to consider the work done by the function as happening in the
context of the supplied parent hash (as distinct from the context of the
hash of the newly arrived block, which is already contained in an outer
span), so this adds noise without conveying extra context.
Instead, use events that occur within the context of the existing spans.
Here the span is added to the body of the `Service::call`
implementation, not to the futures it returns, because the state service
does all of the work synchronously in `call` rather than in the futures
it returns.
The service is skipped as a span field. We could either include or
exclude the request itself. It would be useful, but the request body
can be very large. Instead, we make two spans, one at info level and
one at trace level, and filter that way.
This change is mostly mechanical, with the exception of the changes to the
`tower-batch` middleware. This middleware was adapted from `tower::buffer`,
and the `tower::buffer` code was changed to implement its own bounded queue,
because Tokio 0.3 removed the `mpsc::Sender::poll_send` method. See
ddc64e8d4d
for more context on the Tower changes. To match Tower as closely as possible
in order to be able to upstream `tower-batch`, those changes are copied from
`tower::Buffer` to `tower-batch`.
Some systems have a very small /dev/shm, for example, see:
https://github.com/docker-library/postgres/issues/416
So we should just use the temporary directory on all operating systems.
Also:
* use TempDir to generate the temporary path
* delete the code that we copied from sled
* prefix the temporary path with the state version and network
## Motivation
Prior to this PR we've been using `sled` as our database for storing persistent chain data on the disk between boots. We picked sled over rocksdb to minimize our c++ dependencies despite it being a less mature codebase. The theory was if it worked well enough we'd prefer to have a pure rust codebase, but if we ever ran into problems we knew we could easily swap it out with rocksdb.
Well, we ran into problems. Sled's memory usage was particularly high, and it seemed to be leaking memory. On top of all that, the performance for writes was pretty poor, causing us to become bottle-necked on sled instead of the network.
## Solution
This PR replaces `sled` with `rocksdb`. We've seen a 10x improvement in memory usage out of the box, no more leaking, and much better write performance. With this change writing chain data to disk is no longer a limiting factor in how quickly we can sync the chain.
The code in this pull request has:
- [x] Documentation Comments
- [x] Unit Tests and Property Tests
## Review
@hdevalence
This change explicitly documents cancellation contracts for our Tower services,
and tries to correct a bug in the implementation of the CheckpointVerifier,
which duplicates information from the state service but did not ensure that it
would be kept in sync.
This change has two benefits:
* reduces conflicts with the sled refactor and any replacement
* allows the function to be called independently for testing
`check_contextual_validity` mistakenly used the new block's hash to try
to get the parent block from the state. This caused a panic, because the
new block isn't in the state yet.
Use `StateService::chain` to get the parent block, because we'll be
using `chain` for difficulty adjustment contextual verification anyway.
* Add internal iterator API for accessing relevant chain blocks
* get blocks from all chains in non_finalized state
* Impl FusedIterator for service::Iter
* impl ExactSizedIterator for service::Iter
* let size_hint find heights in side chains
Co-authored-by: teor <teor@riseup.net>
* Add transcript test for requests while state is empty
* Add happy path test for each query once the state is populated
* let populate logic handle out of order blocks
* Add a maximum queued height metric to the finalized state
And rename all the finalized state metrics to contain "finalized".
* Use i32 and -1 instead of Option<Height>
Co-authored-by: Jane Lusby <jlusby42@gmail.com>
Prior to this PR we realized that the RFC had been drafted with the assumption that chains would be ordered from best to worst in `NonFinalizedState`. This assumption was incorrect, since `BTreeSet` only ever orders values in ascending order. This discrepancy was noticed and fixed in the code, but there were still some inconsistencies that needed to be cleaned up.
This PR updates all the incorrect or confusing comments about chain ordering in the RFC and code.
Prior to this PR `memory_state` defined and implemented functionality for three different types, `Chain`, `NonFinalizedState`, and `QueuedBlocks`. Each of these components will need a fair number of unit tests, and I realized that as its currently organized it would be difficult to organize the tests or at a glance figure out which tests are testing which components.
This PR changes the organization of `memory_state` such that each component it exports is defined in its own module. In follow up PRs each module will get its own test module, which will focus exclusively on unit tests for the item defined there-in.
- [Tracking Issue](https://github.com/ZcashFoundation/zebra/issues/1250)
* make service use both finalized and non-finalized state
* Document new functions
* add documentation to sled fns
* cleanup tip fn now that errors are gone
* rename height unwrap fn
## Motivation
While working on the block locator fix PR together with Henry we noticed that we'd accidentally serialized entire transactions in `tx_by_hash`, instead of serializing just the height of the block and the index of the transaction within the block, as described by the original RFC.
## Solution
We've fixed it by adding a `TransactionLocation` new type, which handles the sled format traits. We've removed the sled format impls for `Transaction` to prevent inserting the wrong data in the future. Finally we've bumped the database format to reflect the change in the format on the disk and its incompatibility with previous versions.
Closes#1026
Because of the way that sled uses this parameter, the actual in-memory
size may be much larger. Dialing this down should help avoid high
memory usage.
* Run large checkpoint sync tests in CI
* Improve test child output match error context
* Add a debug_stop_at_height config
* Use stop at height in acceptance tests
And add some restart acceptance tests, to make sure the stop at
height feature works correctly.
## Motivation
The zebra-state service needs to be able to handle duplicate blocks.
## Solution
This implements changes already outlined by [The State
RFC](https://zebra.zfnd.org/dev/rfcs/0005-state-updates.html). We check for
successfully committed blocks first, since interacting with the queued blocks
struct at this point just complicates the implimentation. If the block has not
already been committed we then check if the block has already been queued, if
not we handle the block normally (normally here being the bit we already had
implemented).
## Documentation Changes
- [x] Update the state RFC to match the ways this fix departs from the design
- the main thing is that I switched the order of checking for duplicates
- [x] ~~Add newly added functions to the state rfc~~ Decided not to do this because they're minor getters that don't influence the rest of the design and aren't exposed as part of the API
- [x] Document newly added functions inline
## Testing
## Related Issues
- fixes https://github.com/ZcashFoundation/zebra/issues/1182
- tracking issue https://github.com/ZcashFoundation/zebra/issues/1049
Co-authored-by: teor <teor@riseup.net>
We already use an actor model for the state service, so we get an
ordered sequence of state queries by message-passing. Instead of
performing reads in the futures we return, this commit performs them
synchronously. This means that all sled access is done from the same
task, which
(1) might reduce contention
(2) allows us to avoid using sled transactions when writing to the
state.
Co-authored-by: Jane Lusby <jane@zfnd.org>
Co-authored-by: Jane Lusby <jane@zfnd.org>
Co-authored-by: Deirdre Connolly <deirdre@zfnd.org>
There's no reason to return a pre-Buffer'd service (there's no need for
internal access to the state service, as in zebra-network), but wrapping
it internally removes control of the buffer size from the caller.
This commit begins the process of integrating `zcash_script` with the rest of the system for verifying scripts while syncing the block chain. It does so by adding the necessary support for looking up UTXOs from the state service and implements the first parts of the `script::Verifier` for looking up the necessary UTXOs in the state and then generating the necessary call to `zcash_script` to verify the script itself.
Co-authored-by: teor <teor@riseup.net>
* implement most of the chain functions
* implement fork
* fix outpoint handling in Chain struct
* update expect for work
* split utxo into two sets
* update the Chain definition
* remove allow attribute in zebra-state/lib.rs
* merge ChainSet type into MemoryState
* Add error messages to asserts
* export proptest impls for use in downstream crates
* add testjob for disabled feature in zebra-chain
* try to fix github actions syntax
* add module doc comment
* update RFC for utxos
* add missing header
* working proptest for Chain
* propagate back results over channel
* Start updating RFC to match changes
* implement queued block pruning
* and now it syncs wooo!
* remove empty modules
* setup config for proptests
* re-enable missing_docs lint
* update RFC to match changes in impl
* add documentation
* use more explicit variable names
* Begin work on RFC5 implementation
* I think this is necessary
* holy shit supertrait implemented via subtrait
* implement most of the chain functions
* change to slightly better name
* implement fork
* fix outpoint handling in Chain struct
* update expect for work
* resolve review comment
* split utxo into two sets
* update the Chain definition
* just a little more
* update comment
* Apply suggestions from code review
Co-authored-by: teor <teor@riseup.net>
* apply changes from code review
* remove allow attribute in zebra-state/lib.rs
* Update zebra-state/src/memory_state.rs
Co-authored-by: teor <teor@riseup.net>
* merge ChainSet type into MemoryState
* rename state impl types
* Add error messages to asserts
* checkpoint so I can split off arbitrary changes into a PR
* export proptest impls for use in downstream crates
* add testjob for disabled feature in zebra-chain
* run rustfmt
* try to fix github actions syntax
* differentiate name
* prove that github action tests zebra-chain build without features
* revert change from last commit now that test is running
* remove accidentally introduced newline
* checkpoint
* add module doc comment
* update RFC for utxos
* add missing header
* working proptest for Chain
* apply change from chain impl PR
* setup config for proptests
* Update zebra-chain/src/block/arbitrary.rs
Co-authored-by: teor <teor@riseup.net>
* run rustfmt
Co-authored-by: teor <teor@riseup.net>
* Begin work on RFC5 implementation
* I think this is necessary
* holy shit supertrait implemented via subtrait
* implement most of the chain functions
* change to slightly better name
* implement fork
* fix outpoint handling in Chain struct
* update expect for work
* resolve review comment
* split utxo into two sets
* update the Chain definition
* just a little more
* update comment
* Apply suggestions from code review
Co-authored-by: teor <teor@riseup.net>
* apply changes from code review
* remove allow attribute in zebra-state/lib.rs
* Update zebra-state/src/memory_state.rs
Co-authored-by: teor <teor@riseup.net>
* merge ChainSet type into MemoryState
* rename state impl types
* Add error messages to asserts
* add module doc comment
* update RFC for utxos
* add missing header
Co-authored-by: teor <teor@riseup.net>
* export proptest impls for use in downstream crates
* add testjob for disabled feature in zebra-chain
* run rustfmt
* try to fix github actions syntax
* differentiate name
* prove that github action tests zebra-chain build without features
* revert change from last commit now that test is running
* remove accidentally introduced newline
* Update .github/workflows/ci.yml
Co-authored-by: Deirdre Connolly <deirdre@zfnd.org>
Co-authored-by: Deirdre Connolly <deirdre@zfnd.org>
This addresses at least three pain points:
- we were affected by bugs that were already fixed in git, but not in
the released crate;
- we can use service combinators to transform requests and responses;
- we can use the hedge middleware.
The version in git is still marked as 0.3.1 but these changes will be
part of tower 0.4: https://github.com/tower-rs/tower/issues/431
Using a Buffer with size 1 is a footgun because it allows only one
sender to call poll_ready at a time. This is usually undesirable
because it means that a task or service that calls poll_ready but only
makes a service call later (potentially much later) will block all other
callers.
The previous code filled in block height 0 for a missing coinbase height
in `SledState::commit_finalized`, since the genesis block is the only
block without a coinbase height (because of a mistake when it was
created).
However, @teor2345 noticed that this is incorrect, because we already
parse the genesis block specially and fill in its coinbase height
correctly. So instead, we can .expect it to be present, because we can
assume that all finalized blocks are valid.
The new `StateService` type wraps a `SledState` and a `MemoryState`.
This will allow the sled-related code and the in-memory code to be kept
separate, with the top-level `StateService` making method calls to one
or the other, as appropriate.
This commit removes the existing Service impl for the SledService. This
saves time in refactoring, and the code needs to be rewritten
anyways so there's no loss to deleting it now.
* Remove in-memory state service
* make the config compatible with toml again
* checkpoint commit to see how much I still have to revert
* back to the starting point...
* remove unused dependency
* reorganize error handling a bit
* need to make a new color-eyre release now
* reorder again because I have problems
* remove unnecessary helpers
* revert changes to config loading
* add back missing space
* Switch to released color-eyre version
* add back missing newline again...
* improve error message on unix when terminated by signal
* add context to last few asserts in acceptance tests
* instrument some of the helpers
* remove accidental extra space
* try to make this compile on windows
* reorg platform specific code
* hide on_disk module and fix broken link
Fixes a race condition between the height and hash tree updates, when
they are executed simultaneously with GetDepth. `get_tip` uses the height
tree, but `get` uses the hash tree.
This is the first in a sequence of changes that change the block:: items
to not include Block as a prefix in their name, in accordance with the
Rust API guidelines.
* checkpoint: reject older of duplicate verification requests.
If we get a duplicate block verification request, we should drop the older one
in favor of the newer one, because the older request is likely to have been
canceled. Previously, this code would accept up to four duplicate verification
requests, then fail all subsequent ones.
* sync: add a timeout layer to block requests.
Note that if this timeout is too short, we'll bring down the peer set in a
retry storm.
* sync: restart syncing on error
Restart the syncing process when an error occurs, rather than ignoring it.
Restarting means we discard all tips and start over with a new block locator,
so we can have another chance to "unstuck" ourselves.
* sync: additional debug info
* sync: handle lookahead limit correctly.
Instead of extracting all the completed task results, the previous code pulled
results out until there were fewer tasks than the lookahead limit, then
stopped. This meant that completed tasks could be left until the limit was
exceeded again. Instead, extract all completed results, and use the number of
pending tasks to decide whether to extend the tip or wait for blocks to finish.
* network: add debug instrumentation to retry policy
* sync: instrument the spawned task
* sync: streamline ObtainTips/ExtendTips logic & tracing
This change does three things:
1. It aligns the implementation of ObtainTips and ExtendTips so that they use
the same deduplication method. This means that when debugging we only have one
deduplication algorithm to focus on.
2. It streamlines the tracing output to not include information already
included in spans. Both obtain_tips and extend_tips have their own spans
attached to the events, so it's not necessary to add Scope: prefixes in
messages.
3. It changes the messages to be focused on reporting the actual
events rather than the interpretation of the events (e.g., "got genesis hash in
response" rather than "peer could not extend tip"). The motivation for this
change is that when debugging, the interpretation of events is already known to
be incorrect, in the sense that the mental model of the code (no bug) does not
match its behavior (has bug), so presenting minimally-interpreted events forces
interpretation relative to the actual code.
* sync: hack to work around zcashd behavior
* sync: localize debug statement in extend_tips
* sync: change algorithm to define tips as pairs of hashes.
This is different enough from the existing description that its comments no
longer apply, so I removed them. A further chunk of work is to change the sync
RFC to document this algorithm.
* sync: reduce block timeout
* state: add resource limits for sled
Closes#888
* sync: add a restart timeout constant
* sync: de-pub constants
Instead of creating a block locator all the way back to the genesis
block, only ask for blocks within the reorg limit (99 blocks).
Use the reorg limit as the final locator. (Or if the chain is less
than 99 blocks, use the genesis block.)
Fixes some instances of #818 at very small block heights.
The state service was providing block locators starting at the parent of
the current tip. Instead, include the current tip in the block locator.
Also handle an edge case where we could include the genesis block twice,
if the current tip height was a power of two.
Fixes an instance of #818 where we re-download the current tip.
* Load tracing filter only from config and simplify logic.
* Configure the state storage in the config, not an environment variable.
This also changes the config so that the path is always set rather than being
optional, because Zebra always needs a place to store its config.
* add zebrad acceptance tests
* add custom command test helpers that work with kill
* add and use info event for start and seed commands
* combine conflicting tests into one test case
Co-authored-by: Jane Lusby <jane@zfnd.org>
* Add support for errors in zebra_test::Transcript
* test transcript with an error checker
* switch to option instead of MockError
* update docs
* dont use verifier against ready_and
* cleanup exports and add docs
* handle todos
* fix doctest
* temp: use cleaner error handling example
* add ability to test only for presence of error
We had a brief discussion on discord and it seemed like we had consensus on the
following versioning policy:
* zebrad: match major version to NU version, so we will start by releasing
zebrad 3.0.0;
* zebra-* libraries: start by matching zebrad's version, then increment major
versions of each library as we need to make breaking changes (potentially
faster than the zebrad version, always respecting semver but making no
guarantees about the longevity of major releases).
This commit sets all of the crate versions to 3.0.0-alpha.0 -- the -alpha.0
marks it as a prerelease not subject to perfect adherence to compatibility
guarantees.
Only hash block headers in the lowest-level block index code.
This design has a few benefits:
- failures are obvious, because the hash is not available,
- get_tip() returns a smaller object,
- we avoid re-hashing block headers multiple times.
These efficiency changes may be needed to support chain reorganisations,
multiple tips, and heavy query loads.
Move block header hashing from zebra-consensus to zebra-state.
Handle zebra-state AddBlock errors in zebra-consensus BlockVerifier.
Add unit tests for BlockVerifier state error handling.
Part of #428.
* rename zebra-storage to zebra-state
* Setup initial skeleton for zebra-state
* add test
* Apply suggestions from code review
Co-authored-by: Henry de Valence <hdevalence@hdevalence.ca>
* move shared test vectors to a common crate
Co-authored-by: Jane Lusby <jane@zfnd.org>
Co-authored-by: Henry de Valence <hdevalence@hdevalence.ca>
Henry de Valence
teor
Jane Lusby
Jane Lusby
dependabot[bot]
Alfredo Garcia
Deirdre Connolly
Ramana Venkata
dependabot-preview[bot]