Mercurial > hg
view tests/test-wireproto-clientreactor.py @ 38732:be4984261611
merge: mark file gets as not thread safe (issue5933)
In default installs, this has the effect of disabling the thread-based
worker on Windows when manifesting files in the working directory. My
measurements have shown that with revlog-based repositories, Mercurial
spends a lot of CPU time in revlog code resolving file data. This ends
up incurring a lot of context switching across threads and slows down
`hg update` operations when going from an empty working directory to
the tip of the repo.
On mozilla-unified (246,351 files) on an i7-6700K (4+4 CPUs):
before: 487s wall
after: 360s wall (equivalent to worker.enabled=false)
cpus=2: 379s wall
Even with only 2 threads, the thread pool is still slower.
The introduction of the thread-based worker (02b36e860e0b) states that
it resulted in a "~50%" speedup for `hg sparse --enable-profile` and
`hg sparse --disable-profile`. This disagrees with my measurement
above. I theorize a few reasons for this:
1) Removal of files from the working directory is I/O - not CPU - bound
and should benefit from a thread pool (unless I/O is insanely fast
and the GIL release is near instantaneous). So tests like `hg sparse
--enable-profile` may exercise deletion throughput and aren't good
benchmarks for worker tasks that are CPU heavy.
2) The patch was authored by someone at Facebook. The results were
likely measured against a repository using remotefilelog. And I
believe that revision retrieval during working directory updates with
remotefilelog will often use a remote store, thus being I/O and not
CPU bound. This probably resulted in an overstated performance gain.
Since there appears to be a need to enable the thread-based worker with
some stores, I've made the flagging of file gets as thread safe
configurable. I've made it experimental because I don't want to formalize
a boolean flag for this option and because this attribute is best
captured against the store implementation. But we don't have a proper
store API for this yet. I'd rather cross this bridge later.
It is possible there are revlog-based repositories that do benefit from
a thread-based worker. I didn't do very comprehensive testing. If there
are, we may want to devise a more proper algorithm for whether to use
the thread-based worker, including possibly config options to limit the
number of threads to use. But until I see evidence that justifies
complexity, simplicity wins.
Differential Revision: https://phab.mercurial-scm.org/D3963
author | Gregory Szorc <gregory.szorc@gmail.com> |
---|---|
date | Wed, 18 Jul 2018 09:49:34 -0700 |
parents | deff7cf7eefd |
children | 86b22a4cfab1 |
line wrap: on
line source
from __future__ import absolute_import import unittest from mercurial import ( error, wireprotoframing as framing, ) ffs = framing.makeframefromhumanstring def sendframe(reactor, frame): """Send a frame bytearray to a reactor.""" header = framing.parseheader(frame) payload = frame[framing.FRAME_HEADER_SIZE:] assert len(payload) == header.length return reactor.onframerecv(framing.frame(header.requestid, header.streamid, header.streamflags, header.typeid, header.flags, payload)) class SingleSendTests(unittest.TestCase): """A reactor that can only send once rejects subsequent sends.""" if not getattr(unittest.TestCase, 'assertRaisesRegex', False): # Python 3.7 deprecates the regex*p* version, but 2.7 lacks # the regex version. assertRaisesRegex = (# camelcase-required unittest.TestCase.assertRaisesRegexp) def testbasic(self): reactor = framing.clientreactor(hasmultiplesend=False, buffersends=True) request, action, meta = reactor.callcommand(b'foo', {}) self.assertEqual(request.state, b'pending') self.assertEqual(action, b'noop') action, meta = reactor.flushcommands() self.assertEqual(action, b'sendframes') for frame in meta[b'framegen']: self.assertEqual(request.state, b'sending') self.assertEqual(request.state, b'sent') with self.assertRaisesRegex(error.ProgrammingError, 'cannot issue new commands'): reactor.callcommand(b'foo', {}) with self.assertRaisesRegex(error.ProgrammingError, 'cannot issue new commands'): reactor.callcommand(b'foo', {}) class NoBufferTests(unittest.TestCase): """A reactor without send buffering sends requests immediately.""" def testbasic(self): reactor = framing.clientreactor(hasmultiplesend=True, buffersends=False) request, action, meta = reactor.callcommand(b'command1', {}) self.assertEqual(request.requestid, 1) self.assertEqual(action, b'sendframes') self.assertEqual(request.state, b'pending') for frame in meta[b'framegen']: self.assertEqual(request.state, b'sending') self.assertEqual(request.state, b'sent') action, meta = reactor.flushcommands() self.assertEqual(action, b'noop') # And we can send another command. request, action, meta = reactor.callcommand(b'command2', {}) self.assertEqual(request.requestid, 3) self.assertEqual(action, b'sendframes') for frame in meta[b'framegen']: self.assertEqual(request.state, b'sending') self.assertEqual(request.state, b'sent') class BadFrameRecvTests(unittest.TestCase): if not getattr(unittest.TestCase, 'assertRaisesRegex', False): # Python 3.7 deprecates the regex*p* version, but 2.7 lacks # the regex version. assertRaisesRegex = (# camelcase-required unittest.TestCase.assertRaisesRegexp) def testoddstream(self): reactor = framing.clientreactor() action, meta = sendframe(reactor, ffs(b'1 1 0 1 0 foo')) self.assertEqual(action, b'error') self.assertEqual(meta[b'message'], b'received frame with odd numbered stream ID: 1') def testunknownstream(self): reactor = framing.clientreactor() action, meta = sendframe(reactor, ffs(b'1 0 0 1 0 foo')) self.assertEqual(action, b'error') self.assertEqual(meta[b'message'], b'received frame on unknown stream without beginning ' b'of stream flag set') def testunhandledframetype(self): reactor = framing.clientreactor(buffersends=False) request, action, meta = reactor.callcommand(b'foo', {}) for frame in meta[b'framegen']: pass with self.assertRaisesRegex(error.ProgrammingError, 'unhandled frame type'): sendframe(reactor, ffs(b'1 0 stream-begin text-output 0 foo')) class StreamTests(unittest.TestCase): def testmultipleresponseframes(self): reactor = framing.clientreactor(buffersends=False) request, action, meta = reactor.callcommand(b'foo', {}) self.assertEqual(action, b'sendframes') for f in meta[b'framegen']: pass action, meta = sendframe( reactor, ffs(b'%d 0 stream-begin command-response 0 foo' % request.requestid)) self.assertEqual(action, b'responsedata') action, meta = sendframe( reactor, ffs(b'%d 0 0 command-response eos bar' % request.requestid)) self.assertEqual(action, b'responsedata') if __name__ == '__main__': import silenttestrunner silenttestrunner.main(__name__)