view tests/test-serve.t @ 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

#require serve

  $ hgserve()
  > {
  >    hg serve -a localhost -d --pid-file=hg.pid -E errors.log -v $@ \
  >        | sed -e "s/:$HGPORT1\\([^0-9]\\)/:HGPORT1\1/g" \
  >              -e "s/:$HGPORT2\\([^0-9]\\)/:HGPORT2\1/g" \
  >              -e 's/http:\/\/[^/]*\//http:\/\/localhost\//'
  >    if [ -f hg.pid ]; then
  >        killdaemons.py hg.pid
  >    fi
  >    echo % errors
  >    cat errors.log
  > }

  $ hg init test
  $ cd test
  $ echo '[web]' > .hg/hgrc
  $ echo 'accesslog = access.log' >> .hg/hgrc
  $ echo "port = $HGPORT1" >> .hg/hgrc

Without -v

  $ hg serve -a localhost -p $HGPORT -d --pid-file=hg.pid -E errors.log
  $ cat hg.pid >> "$DAEMON_PIDS"
  $ if [ -f access.log ]; then
  >     echo 'access log created - .hg/hgrc respected'
  > fi
  access log created - .hg/hgrc respected

errors

  $ cat errors.log

With -v

  $ hgserve
  listening at http://localhost/ (bound to *$LOCALIP*:HGPORT1) (glob) (?)
  % errors

With -v and -p HGPORT2

  $ hgserve -p "$HGPORT2"
  listening at http://localhost/ (bound to *$LOCALIP*:HGPORT2) (glob) (?)
  % errors

With -v and -p daytime (should fail because low port)

#if no-root no-windows
  $ KILLQUIETLY=Y
  $ hgserve -p daytime
  abort: cannot start server at 'localhost:13': Permission denied
  abort: child process failed to start
  % errors
  $ KILLQUIETLY=N
#endif

With --prefix foo

  $ hgserve --prefix foo
  listening at http://localhost/foo/ (bound to *$LOCALIP*:HGPORT1) (glob) (?)
  % errors

With --prefix /foo

  $ hgserve --prefix /foo
  listening at http://localhost/foo/ (bound to *$LOCALIP*:HGPORT1) (glob) (?)
  % errors

With --prefix foo/

  $ hgserve --prefix foo/
  listening at http://localhost/foo/ (bound to *$LOCALIP*:HGPORT1) (glob) (?)
  % errors

With --prefix /foo/

  $ hgserve --prefix /foo/
  listening at http://localhost/foo/ (bound to *$LOCALIP*:HGPORT1) (glob) (?)
  % errors

  $ $PYTHON $RUNTESTDIR/killdaemons.py $DAEMON_PIDS

With out of bounds accesses

  $ rm access.log
  $ hg serve -a localhost -p $HGPORT -d --prefix some/dir \
  >    --pid-file=hg.pid -E errors.log
  $ cat hg.pid >> "$DAEMON_PIDS"

  $ hg id http://localhost:$HGPORT/some/dir7
  abort: HTTP Error 404: Not Found
  [255]
  $ hg id http://localhost:$HGPORT/some
  abort: HTTP Error 404: Not Found
  [255]

  $ cat access.log errors.log
  $LOCALIP - - [$LOGDATE$] "GET /some/dir7?cmd=capabilities HTTP/1.1" 404 - (glob)
  $LOCALIP - - [$LOGDATE$] "GET /some?cmd=capabilities HTTP/1.1" 404 - (glob)

  $ cd ..