view tests/test-subrepo-relative-path.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)

Preparing the subrepository 'sub'

  $ hg init sub
  $ echo sub > sub/sub
  $ hg add -R sub
  adding sub/sub
  $ hg commit -R sub -m "sub import"

Preparing the 'main' repo which depends on the subrepo 'sub'

  $ hg init main
  $ echo main > main/main
  $ echo "sub = ../sub" > main/.hgsub
  $ hg clone sub main/sub
  updating to branch default
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg add -R main
  adding main/.hgsub
  adding main/main
  $ hg commit -R main -m "main import"

Cleaning both repositories, just as a clone -U

  $ hg up -C -R sub null
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved
  $ hg up -C -R main null
  0 files updated, 0 files merged, 3 files removed, 0 files unresolved
  $ rm -rf main/sub

hide outer repo
  $ hg init

Serving them both using hgweb

  $ printf '[paths]\n/main = main\nsub = sub\n' > webdir.conf
  $ hg serve --webdir-conf webdir.conf -a localhost -p $HGPORT \
  >    -A /dev/null -E /dev/null --pid-file hg.pid -d
  $ cat hg.pid >> $DAEMON_PIDS

Clone main from hgweb

  $ hg clone "http://user:pass@localhost:$HGPORT/main" cloned
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 3 changes to 3 files
  new changesets fdfeeb3e979e
  updating to branch default
  cloning subrepo sub from http://user@localhost:$HGPORT/sub
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets 863c1745b441
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved

Ensure that subrepos pay attention to default:pushurl

  $ cat > cloned/.hg/hgrc << EOF
  > [paths]
  > default:pushurl = http://user:pass@localhost:$HGPORT/main
  > EOF

  $ hg -R cloned out -S --config paths.default=bogus://invalid
  comparing with http://user:***@localhost:$HGPORT/main
  searching for changes
  no changes found
  comparing with http://user:***@localhost:$HGPORT/sub
  searching for changes
  no changes found
  [1]

TODO: Figure out why, if the password is left out of the default:pushurl URL,
this says "no changes made to subrepo sub since last push".  It looks like from
the original clone command above, the password is getting stripped off, not
just masked out, and that would make the hashed URL different.

  $ hg -R cloned push --config paths.default=bogus://invalid
  pushing to http://user:***@localhost:$HGPORT/main
  pushing subrepo sub to http://user:***@localhost:$HGPORT/sub
  searching for changes
  no changes found
  searching for changes
  no changes found
  abort: HTTP Error 403: ssl required
  [100]

Checking cloned repo ids

  $ hg id -R cloned
  fdfeeb3e979e tip
  $ hg id -R cloned/sub
  863c1745b441 tip

subrepo debug for 'main' clone

  $ hg debugsub -R cloned
  path sub
   source   ../sub
   revision 863c1745b441bd97a8c4a096e87793073f4fb215

Test sharing with a remote URL reference

  $ hg init absolute_subrepo
  $ cd absolute_subrepo
  $ echo foo > foo.txt
  $ hg ci -Am 'initial commit'
  adding foo.txt
  $ echo "sub = http://localhost:$HGPORT/sub" > .hgsub
  $ hg ci -Am 'add absolute subrepo'
  adding .hgsub
  $ cd ..

Clone pooling works for local clones with a remote subrepo reference.  The
subrepo is cloned to the pool and shared from there, so that all clones will
share the same subrepo.

  $ hg --config extensions.share= --config share.pool=$TESTTMP/pool \
  >    clone absolute_subrepo cloned_from_abs
  (sharing from new pooled repository 8d6a2f1e993b34b6557de0042cfe825ae12a8dae)
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 2 changesets with 3 changes to 3 files
  new changesets 8d6a2f1e993b:* (glob)
  searching for changes
  no changes found
  updating working directory
  cloning subrepo sub from http://localhost:$HGPORT/sub
  (sharing from new pooled repository 863c1745b441bd97a8c4a096e87793073f4fb215)
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets 863c1745b441
  searching for changes
  no changes found
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved

Vanilla sharing with a subrepo remote path reference will clone the subrepo.
Each share of these top level repos will end up with independent subrepo copies
(potentially leaving the shared parent with dangling cset references).

  $ hg --config extensions.share= share absolute_subrepo shared_from_abs
  updating working directory
  cloning subrepo sub from http://localhost:$HGPORT/sub
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets 863c1745b441
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ hg --config extensions.share= share -U absolute_subrepo shared_from_abs2
  $ hg -R shared_from_abs2 update -r tip
  cloning subrepo sub from http://localhost:$HGPORT/sub
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets 863c1745b441
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved

A parent repo without its subrepo available locally can be shared if the
subrepo is referenced by absolute path.

  $ hg clone -U absolute_subrepo cloned_null_from_abs
  $ hg --config extensions.share= share cloned_null_from_abs shared_from_null_abs
  updating working directory
  cloning subrepo sub from http://localhost:$HGPORT/sub
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets 863c1745b441
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ killdaemons.py

subrepo paths with ssh urls

  $ hg clone ssh://user@dummy/cloned sshclone
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 3 changes to 3 files
  new changesets fdfeeb3e979e
  updating to branch default
  cloning subrepo sub from ssh://user@dummy/sub
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets 863c1745b441
  3 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ hg -R sshclone push ssh://user@dummy/`pwd`/cloned
  pushing to ssh://user@dummy/$TESTTMP/cloned
  pushing subrepo sub to ssh://user@dummy/$TESTTMP/sub
  searching for changes
  no changes found
  searching for changes
  no changes found
  [1]

  $ cat dummylog
  Got arguments 1:user@dummy 2:hg -R cloned serve --stdio
  Got arguments 1:user@dummy 2:hg -R sub serve --stdio
  Got arguments 1:user@dummy 2:hg -R $TESTTMP/cloned serve --stdio
  Got arguments 1:user@dummy 2:hg -R $TESTTMP/sub serve --stdio