view tests/test-fastannotate-protocol.t @ 51681:522b4d729e89

mmap: populate the mapping by default

Without pre-population, accessing all data through a mmap can result in many
pagefault, reducing performance significantly. If the mmap is prepopulated, the
performance can no longer get slower than a full read.
(See benchmark number below)

In some cases were very few data is read, prepopulating can be overkill and
slower than populating on access (through page fault). So that behavior can be
controlled when the caller can pre-determine the best behavior.
(See benchmark number below)

In addition, testing with populating in a secondary thread yield great result
combining the best of each approach. This might be implemented in later
changesets.

In all cases, using mmap has a great effect on memory usage when many processes
run in parallel on the same machine.

### Benchmarks

# What did I run

A couple of month back I ran a large benchmark campaign to assess the impact of
various approach for using mmap with the revlog (and other files), it
highlighted a few benchmarks that capture the impact of the changes well. So to
validate this change I checked the following:

- log command displaying various revisions
  (read the changelog index)
- log command displaying the patch of listed revisions
  (read the changelog index, the manifest index and a few files indexes)
- unbundling a few revisions
  (read and write changelog, manifest and few files indexes, and walk the graph
  to update some cache)
- pushing a few revisions
  (read and write changelog, manifest and few files indexes, walk the graph to
  update some cache, performs various accesses locally and remotely during
  discovery)

Benchmarks were run using the default module policy (c+py) and the rust one. No
significant difference were found between the two implementation, so we will
present result using the default policy (unless otherwise specified).

I ran them on a few repositories :

- mercurial: a "public changeset only" copy of mercurial from 2018-08-01 using
  zstd compression and sparse-revlog

- pypy: a copy of pypy from 2018-08-01 using zstd compression and sparse-revlog

- netbeans: a copy of netbeans from 2018-08-01 using zstd compression and
  sparse-revlog

- mozilla-try: a copy of mozilla-try from 2019-02-18 using zstd compression and
  sparse-revlog

- mozilla-try persistent-nodemap: Same as the above but with a persistent
  nodemap. Used for the log --patch benchmark only

# Results

For the smaller repositories (mercurial, pypy), the impact of mmap is almost
imperceptible, other cost dominating the operation. The impact of prepopulating
is undiscernible in the benchmark we ran.

For larger repositories the benchmark support explanation given above:

On netbeans, the log can be about 1% faster without repopulation (for a
difference < 100ms) but unbundle becomes a bit slower, even when small.

### data-env-vars.name                             = netbeans-2018-08-01-zstd-sparse-revlog
  # benchmark.name                                 = hg.command.unbundle
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
  # benchmark.variants.source                      = unbundle
  # benchmark.variants.verbosity                   = quiet
with-populate: 0.240157
no-populate:   0.265087  (+10.38%, +0.02)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.459518
no-populate:   1.481290   (+1.49%, +0.02)
 ## benchmark.name                                 = hg.command.push
  # benchmark.variants.explicit-rev                = none
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.protocol                    = ssh
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 0.771919
no-populate:   0.792025   (+2.60%, +0.02)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.459518
no-populate:   1.481290   (+1.49%, +0.02)


For mozilla-try, the "slow down" from pre-populate for small `hg log` is more
visible, but still small in absolute time. (using rust value for the persistent
nodemap value to be relevant).

### data-env-vars.name            = mozilla-try-2019-02-18-ds2-pnm
  # benchmark.name                = hg.command.log
  # bin-env-vars.hg.flavor        = rust
  # benchmark.variants.patch      = yes
  # benchmark.variants.limit-rev  = 1
with-populate: 0.237813
no-populate:   0.229452   (-3.52%, -0.01)
  # benchmark.variants.limit-rev  = 10
  # benchmark.variants.patch      = yes
with-populate: 1.213578
no-populate:   1.205189
### data-env-vars.name            = mozilla-try-2019-02-18-zstd-sparse-revlog
  # benchmark.variants.limit-rev  = 1000
  # benchmark.variants.patch      = no
  # benchmark.variants.rev        = tip
with-populate: 0.198607
no-populate:   0.195038   (-1.80%, -0.00)

However pre-populating provide a significant boost on more complex operations
like unbundle or push:

### data-env-vars.name                             = mozilla-try-2019-02-18-zstd-sparse-revlog
  # benchmark.name                                 = hg.command.push
  # benchmark.variants.explicit-rev                = none
  # benchmark.variants.issue6528                   = disabled
  # benchmark.variants.protocol                    = ssh
  # benchmark.variants.reuse-external-delta-parent = yes
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 4.798632
no-populate:   4.953295   (+3.22%, +0.15)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 4.903618
no-populate:   5.014963   (+2.27%, +0.11)
 ## benchmark.name                                 = hg.command.unbundle
  # benchmark.variants.revs                        = any-1-extra-rev
with-populate: 1.423411
no-populate:   1.585365  (+11.38%, +0.16)
  # benchmark.variants.revs                        = any-100-extra-rev
with-populate: 1.537909
no-populate:   1.688489   (+9.79%, +0.15)

  $ cat >> $HGRCPATH << EOF
  > [extensions]
  > fastannotate=
  > [fastannotate]
  > mainbranch=@
  > EOF

setup the server repo

  $ hg init repo-server
  $ cd repo-server
  $ cat >> .hg/hgrc << EOF
  > [fastannotate]
  > server=1
  > EOF
  $ for i in 1 2 3 4; do
  >   echo $i >> a
  >   hg commit -A -m $i a
  > done
  $ [ -d .hg/fastannotate ]
  [1]
  $ hg bookmark @
  $ cd ..

setup the local repo

  $ hg clone 'ssh://user@dummy/repo-server' repo-local -q
  $ cd repo-local
  $ cat >> .hg/hgrc << EOF
  > [fastannotate]
  > client=1
  > clientfetchthreshold=0
  > EOF
  $ [ -d .hg/fastannotate ]
  [1]
  $ hg fastannotate a --debug
  running * (glob)
  sending hello command
  sending between command
  remote: * (glob) (?)
  remote: capabilities: * (glob)
  remote: * (glob) (?)
  sending protocaps command
  fastannotate: requesting 1 files
  sending getannotate command
  fastannotate: writing 112 bytes to fastannotate/default/a.l
  fastannotate: writing 94 bytes to fastannotate/default/a.m
  fastannotate: a: using fast path (resolved fctx: True)
  0: 1
  1: 2
  2: 3
  3: 4

the cache could be reused and no download is necessary

  $ hg fastannotate a --debug
  fastannotate: a: using fast path (resolved fctx: True)
  0: 1
  1: 2
  2: 3
  3: 4

if the client agrees where the head of the master branch is, no re-download
happens even if the client has more commits

  $ echo 5 >> a
  $ hg commit -m 5
  $ hg bookmark -r 3 @ -f
  $ hg fastannotate a --debug
  0: 1
  1: 2
  2: 3
  3: 4
  4: 5

if the client has a different "@" (head of the master branch) and "@" is ahead
of the server, the server can detect things are unchanged and does not return
full contents (not that there is no "writing ... to fastannotate"), but the
client can also build things up on its own (causing diverge)

  $ hg bookmark -r 4 @ -f
  $ hg fastannotate a --debug
  running * (glob)
  sending hello command
  sending between command
  remote: * (glob) (?)
  remote: capabilities: * (glob)
  remote: * (glob) (?)
  sending protocaps command
  fastannotate: requesting 1 files
  sending getannotate command
  fastannotate: a: 1 new changesets in the main branch
  0: 1
  1: 2
  2: 3
  3: 4
  4: 5

if the client has a different "@" which is behind the server. no download is
necessary

  $ hg fastannotate a --debug --config fastannotate.mainbranch=2
  fastannotate: a: using fast path (resolved fctx: True)
  0: 1
  1: 2
  2: 3
  3: 4
  4: 5

define fastannotate on-disk paths

  $ p1=.hg/fastannotate/default
  $ p2=../repo-server/.hg/fastannotate/default

revert bookmark change so the client is behind the server

  $ hg bookmark -r 2 @ -f

in the "fctx" mode with the "annotate" command, the client also downloads the
cache. but not in the (default) "fastannotate" mode.

  $ rm $p1/a.l $p1/a.m
  $ hg annotate a --debug | grep 'fastannotate: writing'
  [1]
  $ hg annotate a --config fastannotate.modes=fctx --debug | grep 'fastannotate: writing' | sort
  fastannotate: writing 112 bytes to fastannotate/default/a.l
  fastannotate: writing 94 bytes to fastannotate/default/a.m

the fastannotate cache (built server-side, downloaded client-side) in two repos
have the same content (because the client downloads from the server)

  $ diff $p1/a.l $p2/a.l
  $ diff $p1/a.m $p2/a.m

in the "fctx" mode, the client could also build the cache locally

  $ hg annotate a --config fastannotate.modes=fctx --debug --config fastannotate.mainbranch=4 | grep fastannotate
  fastannotate: requesting 1 files
  fastannotate: a: 1 new changesets in the main branch

the server would rebuild broken cache automatically

  $ cp $p2/a.m $p2/a.m.bak
  $ echo BROKEN1 > $p1/a.m
  $ echo BROKEN2 > $p2/a.m
  $ hg fastannotate a --debug | grep 'fastannotate: writing' | sort
  fastannotate: writing 112 bytes to fastannotate/default/a.l
  fastannotate: writing 94 bytes to fastannotate/default/a.m
  $ diff $p1/a.m $p2/a.m
  $ diff $p2/a.m $p2/a.m.bak

use the "debugbuildannotatecache" command to build annotate cache

  $ rm -rf $p1 $p2
  $ hg --cwd ../repo-server debugbuildannotatecache a --debug
  fastannotate: a: 4 new changesets in the main branch
  $ hg --cwd ../repo-local debugbuildannotatecache a --debug
  running * (glob)
  sending hello command
  sending between command
  remote: * (glob) (?)
  remote: capabilities: * (glob)
  remote: * (glob) (?)
  sending protocaps command
  fastannotate: requesting 1 files
  sending getannotate command
  fastannotate: writing * (glob)
  fastannotate: writing * (glob)
  $ diff $p1/a.l $p2/a.l
  $ diff $p1/a.m $p2/a.m

with the clientfetchthreshold config option, the client can build up the cache
without downloading from the server

  $ rm -rf $p1
  $ hg fastannotate a --debug --config fastannotate.clientfetchthreshold=10
  fastannotate: a: 3 new changesets in the main branch
  0: 1
  1: 2
  2: 3
  3: 4
  4: 5

if the fastannotate directory is not writable, the fctx mode still works

  $ rm -rf $p1
  $ touch $p1
  $ hg annotate a --debug --traceback --config fastannotate.modes=fctx
  fastannotate: a: cache broken and deleted
  fastannotate: prefetch failed: * (glob)
  fastannotate: a: cache broken and deleted
  fastannotate: falling back to the vanilla annotate: * (glob)
  0: 1
  1: 2
  2: 3
  3: 4
  4: 5

with serverbuildondemand=False, the server will not build anything

  $ cat >> ../repo-server/.hg/hgrc <<EOF
  > [fastannotate]
  > serverbuildondemand=False
  > EOF
  $ rm -rf $p1 $p2
  $ hg fastannotate a --debug | grep 'fastannotate: writing'
  [1]