38fd85c676a072ebf256e806beda9d7533790baa http: replace WorkQueue and threads handling for ThreadPool (furszy)
c323f882ed3841401edee90ab5261d68215ab316 fuzz: add test case for threadpool (TheCharlatan)
c528dd5f8ccc3955b00bdba869f0a774efa97fe1 util: introduce general purpose thread pool (furszy)
6354b4fd7fe819eb13274b212e426a7d10ca75d3 tests: log node JSON-RPC errors during test setup (furszy)
45930a79412dc45f9d391cd7689d029fa4f0189e http-server: guard against crashes from unhandled exceptions (furszy)
Pull request description:
This has been a recent discovery; the general thread pool class created for #26966, cleanly
integrates into the HTTP server. It simplifies init, shutdown and requests execution logic.
Replacing code that was never unit tested for code that is properly unit and fuzz tested.
Although our functional test framework extensively uses this RPC interface (that’s how
we’ve been ensuring its correct behavior so far - which is not the best).
This clearly separates the responsibilities:
The HTTP server now focuses solely on receiving and dispatching requests, while ThreadPool handles
concurrency, queuing, and execution.
This will also allows us to experiment with further performance improvements at the task queuing and
execution level, such as a lock-free structure or task prioritization or any other implementation detail
like coroutines in the future, without having to deal with HTTP code that lives on a different layer.
Note:
The rationale behind introducing the ThreadPool first is to be able to easily cherry-pick it across different
working paths. Some of the ones that are benefited from it are #26966 for the parallelization of the indexes
initial sync, #31132 for the parallelization of the inputs fetching procedure, #32061 for the libevent replacement,
the kernel API #30595 (https://github.com/bitcoin/bitcoin/pull/30595#discussion_r2413702370) to avoid blocking validation among others use cases not publicly available.
Note 2:
I could have created a wrapper around the existing code and replaced the `WorkQueue` in a subsequent
commit, but it didn’t seem worth the extra commits and review effort. The `ThreadPool` implements
essentially the same functionality in a more modern and cleaner way.
ACKs for top commit:
Eunovo:
ReACK 38fd85c676
sedited:
Re-ACK 38fd85c676a072ebf256e806beda9d7533790baa
pinheadmz:
ACK 38fd85c676a072ebf256e806beda9d7533790baa
Tree-SHA512: a0330e54ed504330ca874c42d4e318a909f548b2fb9ac46db8badf5935b9eec47dc4ed503d1b6f98574418e3473420ea45f60498be05545c4325cfa89dcca689
Functional tests
Writing Functional Tests
Example test
The file test/functional/example_test.py is a heavily commented example of a test case that uses both the RPC and P2P interfaces. If you are writing your first test, copy that file and modify to fit your needs.
Coverage
Assuming the build directory is build,
running build/test/functional/test_runner.py with the --coverage argument tracks which RPCs are
called by the tests and prints a report of uncovered RPCs in the summary. This
can be used (along with the --extended argument) to find out which RPCs we
don't have test cases for.
Style guidelines
- Where possible, try to adhere to PEP-8 guidelines
- Use a python linter like flake8 before submitting PRs to catch common style nits (eg trailing whitespace, unused imports, etc)
- The oldest supported Python version is specified in doc/dependencies.md. Consider using pyenv, which checks .python-version, to prevent accidentally introducing modern syntax from an unsupported Python version. The CI linter job also checks this, but possibly not in all cases.
- See the python lint script that checks for violations that could lead to bugs and issues in the test code.
- Use type hints in your code to improve code readability and to detect possible bugs earlier.
- Avoid wildcard imports.
- If more than one name from a module is needed, use lexicographically sorted multi-line imports in order to reduce the possibility of potential merge conflicts.
- Use a module-level docstring to describe what the test is testing, and how it is testing it.
- When subclassing the BitcoinTestFramework, place overrides for the
set_test_params(),add_options()andsetup_xxxx()methods at the top of the subclass, then locally-defined helper methods, then therun_test()method. - Use
f'{x}'for string formatting in preference to'{}'.format(x)or'%s' % x. - Use
platform.system()for detecting the running operating system andos.nameto check whether it's a POSIX system (see also theskip_if_platform_not_{linux,posix}methods in theBitcoinTestFrameworkclass, which can be used to skip a whole test depending on the platform).
Naming guidelines
- Name the test
<area>_test.py, where area can be one of the following:featurefor tests for full features that aren't wallet/mining/mempool, egfeature_rbf.pyinterfacefor tests for other interfaces (REST, ZMQ, etc), eginterface_rest.pymempoolfor tests for mempool behaviour, egmempool_reorg.pyminingfor tests for mining features, egmining_prioritisetransaction.pyp2pfor tests that explicitly test the p2p interface, egp2p_disconnect_ban.pyrpcfor tests for individual RPC methods or features, egrpc_listtransactions.pytoolfor tests for tools, egtool_wallet.pywalletfor tests for wallet features, egwallet_keypool.py
- Use an underscore to separate words
- exception: for tests for specific RPCs or command line options which don't include underscores, name the test after the exact RPC or argument name, eg
rpc_decodescript.py, notrpc_decode_script.py
- exception: for tests for specific RPCs or command line options which don't include underscores, name the test after the exact RPC or argument name, eg
- Don't use the redundant word
testin the name, eginterface_zmq.py, notinterface_zmq_test.py
General test-writing advice
- Instead of inline comments or no test documentation at all, log the comments to the test log, e.g.
self.log.info('Create enough transactions to fill a block'). Logs make the test code easier to read and the test logic easier to debug. - Set
self.num_nodesto the minimum number of nodes necessary for the test. Having additional unrequired nodes adds to the execution time of the test as well as memory/CPU/disk requirements (which is important when running tests in parallel). - Avoid stop-starting the nodes multiple times during the test if possible. A stop-start takes several seconds, so doing it several times blows up the runtime of the test.
- Set the
self.setup_clean_chainvariable inset_test_params()toTrueto initialize an empty blockchain and start from the Genesis block, rather than load a premined blockchain from cache with the default value ofFalse. The cached data directories contain a 200-block pre-mined blockchain with the spendable mining rewards being split between four nodes. Each node has 25 mature block subsidies (25x50=1250 BTC) in its wallet. Using them is much more efficient than mining blocks in your test. - When calling RPCs with lots of arguments, consider using named keyword arguments instead of positional arguments to make the intent of the call clear to readers.
- Many of the core test framework classes such as
CBlockandCTransactiondon't allow new attributes to be added to their objects at runtime like typical Python objects allow. This helps prevent unpredictable side effects from typographical errors or usage of the objects outside of their intended purpose.
RPC and P2P definitions
Test writers may find it helpful to refer to the definitions for the RPC and P2P messages. These can be found in the following source files:
/src/rpc/*for RPCs/src/wallet/rpc*for wallet RPCsProcessMessage()in/src/net_processing.cppfor parsing P2P messages
Using the P2P interface
-
P2Ps can be used to test specific P2P protocol behavior. p2p.py contains test framework p2p objects and messages.py contains all the definitions for objects passed over the network (CBlock,CTransaction, etc, along with the network-level wrappers for them,msg_block,msg_tx, etc). -
P2P tests have two threads. One thread handles all network communication with the bitcoind(s) being tested in a callback-based event loop; the other implements the test logic.
-
P2PConnectionis the class used to connect to a bitcoind.P2PInterfacecontains the higher level logic for processing P2P payloads and connecting to the Bitcoin Core node application logic. For custom behaviour, subclass the P2PInterface object and override the callback methods.
P2PConnections can be used as such:
p2p_conn = node.add_p2p_connection(P2PInterface())
p2p_conn.send_and_ping(msg)
They can also be referenced by indexing into a TestNode's p2ps list, which
contains the list of test framework p2p objects connected to itself
(it does not include any TestNodes):
node.p2ps[0].sync_with_ping()
More examples can be found in p2p_unrequested_blocks.py, p2p_compactblocks.py.
Prototyping tests
The TestShell class exposes the BitcoinTestFramework
functionality to interactive Python3 environments and can be used to prototype
tests. This may be especially useful in a REPL environment with session logging
utilities, such as
IPython.
The logs of such interactive sessions can later be adapted into permanent test
cases.
Test framework modules
The following are useful modules for test developers. They are located in test/functional/test_framework/.
authproxy.py
Taken from the python-bitcoinrpc repository.
test_framework.py
Base class for functional tests.
util.py
Generally useful functions.
p2p.py
Test objects for interacting with a bitcoind node over the p2p interface.
script.py
Utilities for manipulating transaction scripts (originally from python-bitcoinlib)
key.py
Test-only secp256k1 elliptic curve implementation
blocktools.py
Helper functions for creating blocks and transactions.
Benchmarking with perf
An easy way to profile node performance during functional tests is provided
for Linux platforms using perf.
Perf will sample the running node and will generate profile data in the node's
datadir. The profile data can then be presented using perf report or a graphical
tool like hotspot.
There are two ways of invoking perf: one is to use the --perf flag when
running tests, which will profile each node during the entire test run: perf
begins to profile when the node starts and ends when it shuts down. The other
way is the use the profile_with_perf context manager, e.g.
with node.profile_with_perf("send-big-msgs"):
# Perform activity on the node you're interested in profiling, e.g.:
for _ in range(10000):
node.p2ps[0].send_without_ping(some_large_message)
To see useful textual output, run
perf report -i /path/to/datadir/send-big-msgs.perf.data.xxxx --stdio | c++filt | less
See also:
- Installing perf
- Perf examples
- Hotspot: a GUI for perf output analysis