tests/test-largefiles-cache.t

localrepo: experimental support for non-zlib revlog compression

The final part of integrating the compression manager APIs into
revlog storage is the plumbing for repositories to advertise they
are using non-zlib storage and for revlogs to instantiate a non-zlib
compression engine.

The main intent of the compression manager work was to zstd all
of the things. Adding zstd to revlogs has proved to be more involved
than other places because revlogs are... special. Very small inputs
and the use of delta chains (which are themselves a form of
compression) are a completely different use case from streaming
compression, which bundles and the wire protocol employ. I've
conducted numerous experiments with zstd in revlogs and have yet
to formalize compression settings and a storage architecture that
I'm confident I won't regret later. In other words, I'm not yet
ready to commit to a new mechanism for using zstd - or any other
compression format - in revlogs.

That being said, having some support for zstd (and other compression
formats) in revlogs in core is beneficial. It can allow others to
conduct experiments.

This patch introduces *highly experimental* support for non-zlib
compression formats in revlogs. Introduced is a config option to
control which compression engine to use. Also introduced is a namespace
of "exp-compression-*" requirements to denote support for non-zlib
compression in revlogs. I've prefixed the namespace with "exp-"
(short for "experimental") because I'm not confident of the
requirements "schema" and in no way want to give the illusion of
supporting these requirements in the future. I fully intend to drop
support for these requirements once we figure out what we're doing
with zstd in revlogs.

A good portion of the patch is teaching the requirements system
about registered compression engines and passing the requested
compression engine as an opener option so revlogs can instantiate
the proper compression engine for new operations.

That's a verbose way of saying "we can now use zstd in revlogs!"

On an `hg pull` conversion of the mozilla-unified repo with no extra
redelta settings (like aggressivemergedeltas), we can see the impact
of zstd vs zlib in revlogs:

$ hg perfrevlogchunks -c
! chunk
! wall 2.032052 comb 2.040000 user 1.990000 sys 0.050000 (best of 5)
! wall 1.866360 comb 1.860000 user 1.820000 sys 0.040000 (best of 6)

! chunk batch
! wall 1.877261 comb 1.870000 user 1.860000 sys 0.010000 (best of 6)
! wall 1.705410 comb 1.710000 user 1.690000 sys 0.020000 (best of 6)

$ hg perfrevlogchunks -m
! chunk
! wall 2.721427 comb 2.720000 user 2.640000 sys 0.080000 (best of 4)
! wall 2.035076 comb 2.030000 user 1.950000 sys 0.080000 (best of 5)

! chunk batch
! wall 2.614561 comb 2.620000 user 2.580000 sys 0.040000 (best of 4)
! wall 1.910252 comb 1.910000 user 1.880000 sys 0.030000 (best of 6)

$ hg perfrevlog -c -d 1
! wall 4.812885 comb 4.820000 user 4.800000 sys 0.020000 (best of 3)
! wall 4.699621 comb 4.710000 user 4.700000 sys 0.010000 (best of 3)

$ hg perfrevlog -m -d 1000
! wall 34.252800 comb 34.250000 user 33.730000 sys 0.520000 (best of 3)
! wall 24.094999 comb 24.090000 user 23.320000 sys 0.770000 (best of 3)

Only modest wins for the changelog. But manifest reading is
significantly faster. What's going on?

One reason might be data volume. zstd decompresses faster. So given
more bytes, it will put more distance between it and zlib.

Another reason is size. In the current design, zstd revlogs are
*larger*:

debugcreatestreamclonebundle (size in bytes)
zlib: 1,638,852,492
zstd: 1,680,601,332

I haven't investigated this fully, but I reckon a significant cause of
larger revlogs is that the zstd frame/header has more bytes than
zlib's. For very small inputs or data that doesn't compress well, we'll
tend to store more uncompressed chunks than with zlib (because the
compressed size isn't smaller than original). This will make revlog
reading faster because it is doing less decompression.

Moving on to bundle performance:

$ hg bundle -a -t none-v2 (total CPU time)
zlib: 102.79s
zstd:  97.75s

So, marginal CPU decrease for reading all chunks in all revlogs
(this is somewhat disappointing).

$ hg bundle -a -t <engine>-v2 (total CPU time)
zlib: 191.59s
zstd: 115.36s

This last test effectively measures the difference between zlib->zlib
and zstd->zstd for revlogs to bundle. This is a rough approximation of
what a server does during `hg clone`.

There are some promising results for zstd. But not enough for me to
feel comfortable advertising it to users. We'll get there...

Create user cache directory

  $ USERCACHE=`pwd`/cache; export USERCACHE
  $ cat <<EOF >> ${HGRCPATH}
  > [extensions]
  > hgext.largefiles=
  > [largefiles]
  > usercache=${USERCACHE}
  > EOF
  $ mkdir -p ${USERCACHE}

Create source repo, and commit adding largefile.

  $ hg init src
  $ cd src
  $ echo large > large
  $ hg add --large large
  $ hg commit -m 'add largefile'
  $ hg rm large
  $ hg commit -m 'branchhead without largefile' large
  $ hg up -qr 0
  $ rm large
  $ echo "0000000000000000000000000000000000000000" > .hglf/large
  $ hg commit -m 'commit missing file with corrupt standin' large
  abort: large: file not found!
  [255]
  $ hg up -Cqr 0
  $ cd ..

Discard all cached largefiles in USERCACHE

  $ rm -rf ${USERCACHE}

Create mirror repo, and pull from source without largefile:
"pull" is used instead of "clone" for suppression of (1) updating to
tip (= caching largefile from source repo), and (2) recording source
repo as "default" path in .hg/hgrc.

  $ hg init mirror
  $ cd mirror
  $ hg pull ../src
  pulling from ../src
  requesting all changes
  adding changesets
  adding manifests
  adding file changes
  added 2 changesets with 1 changes to 1 files
  (run 'hg update' to get a working copy)

Update working directory to "tip", which requires largefile("large"),
but there is no cache file for it.  So, hg must treat it as
"missing"(!) file.

  $ hg update -r0
  getting changed largefiles
  large: largefile 7f7097b041ccf68cc5561e9600da4655d21c6d18 not available from file:/*/$TESTTMP/mirror (glob)
  0 largefiles updated, 0 removed
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg status
  ! large

Update working directory to null: this cleanup .hg/largefiles/dirstate

  $ hg update null
  getting changed largefiles
  0 largefiles updated, 0 removed
  0 files updated, 0 files merged, 1 files removed, 0 files unresolved

Update working directory to tip, again.

  $ hg update -r0
  getting changed largefiles
  large: largefile 7f7097b041ccf68cc5561e9600da4655d21c6d18 not available from file:/*/$TESTTMP/mirror (glob)
  0 largefiles updated, 0 removed
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ hg status
  ! large
  $ cd ..

Verify that largefiles from pulled branchheads are fetched, also to an empty repo

  $ hg init mirror2
  $ hg -R mirror2 pull src -r0
  pulling from src
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  (run 'hg update' to get a working copy)

#if unix-permissions

Portable way to print file permissions:

  $ cat > ls-l.py <<EOF
  > #!/usr/bin/env python
  > import sys, os
  > path = sys.argv[1]
  > print '%03o' % (os.lstat(path).st_mode & 0777)
  > EOF
  $ chmod +x ls-l.py

Test that files in .hg/largefiles inherit mode from .hg/store, not
from file in working copy:

  $ cd src
  $ chmod 750 .hg/store
  $ chmod 660 large
  $ echo change >> large
  $ hg commit -m change
  created new head
  $ ../ls-l.py .hg/largefiles/e151b474069de4ca6898f67ce2f2a7263adf8fea
  640

Test permission of with files in .hg/largefiles created by update:

  $ cd ../mirror
  $ rm -r "$USERCACHE" .hg/largefiles # avoid links
  $ chmod 750 .hg/store
  $ hg pull ../src --update -q
  $ ../ls-l.py .hg/largefiles/e151b474069de4ca6898f67ce2f2a7263adf8fea
  640

Test permission of files created by push:

  $ hg serve -R ../src -d -p $HGPORT --pid-file hg.pid \
  >          --config "web.allow_push=*" --config web.push_ssl=no
  $ cat hg.pid >> $DAEMON_PIDS

  $ echo change >> large
  $ hg commit -m change

  $ rm -r "$USERCACHE"

  $ hg push -q http://localhost:$HGPORT/

  $ ../ls-l.py ../src/.hg/largefiles/b734e14a0971e370408ab9bce8d56d8485e368a9
  640

  $ cd ..

#endif

Test issue 4053 (remove --after on a deleted, uncommitted file shouldn't say
it is missing, but a remove on a nonexistent unknown file still should.  Same
for a forget.)

  $ cd src
  $ touch x
  $ hg add x
  $ mv x y
  $ hg remove -A x y ENOENT
  ENOENT: * (glob)
  not removing y: file is untracked
  [1]
  $ hg add y
  $ mv y z
  $ hg forget y z ENOENT
  ENOENT: * (glob)
  not removing z: file is already untracked
  [1]

Largefiles are accessible from the share's store
  $ cd ..
  $ hg share -q src share_dst --config extensions.share=
  $ hg -R share_dst update -r0
  getting changed largefiles
  1 largefiles updated, 0 removed
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ echo modified > share_dst/large
  $ hg -R share_dst ci -m modified
  created new head

Only dirstate is in the local store for the share, and the largefile is in the
share source's local store.  Avoid the extra largefiles added in the unix
conditional above.
  $ hash=`hg -R share_dst cat share_dst/.hglf/large`
  $ echo $hash
  e2fb5f2139d086ded2cb600d5a91a196e76bf020

  $ find share_dst/.hg/largefiles/* | sort
  share_dst/.hg/largefiles/dirstate

  $ find src/.hg/largefiles/* | egrep "(dirstate|$hash)" | sort
  src/.hg/largefiles/dirstate
  src/.hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020

Verify that backwards compatibility is maintained for old storage layout
  $ mv src/.hg/largefiles/$hash share_dst/.hg/largefiles
  $ hg verify --quiet --lfa -R share_dst --config largefiles.usercache=

Inject corruption into the largefiles store and see how update handles that:

  $ cd src
  $ hg up -qC tip
  $ cat large
  modified
  $ rm large
  $ cat .hglf/large
  e2fb5f2139d086ded2cb600d5a91a196e76bf020
  $ mv .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020 ..
  $ echo corruption > .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020
  $ hg up -C
  getting changed largefiles
  large: data corruption in $TESTTMP/src/.hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020 with hash 6a7bb2556144babe3899b25e5428123735bb1e27 (glob)
  0 largefiles updated, 0 removed
  0 files updated, 0 files merged, 0 files removed, 0 files unresolved
  [12] other heads for branch "default" (re)
  $ hg st
  ! large
  ? z
  $ rm .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020

#if serve

Test coverage of error handling from putlfile:

  $ mkdir $TESTTMP/mirrorcache
  $ hg serve -R ../mirror -d -p $HGPORT1 --pid-file hg.pid --config largefiles.usercache=$TESTTMP/mirrorcache
  $ cat hg.pid >> $DAEMON_PIDS

  $ hg push http://localhost:$HGPORT1 -f --config files.usercache=nocache
  pushing to http://localhost:$HGPORT1/
  searching for changes
  abort: remotestore: could not open file $TESTTMP/src/.hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020: HTTP Error 403: ssl required
  [255]

  $ rm .hg/largefiles/e2fb5f2139d086ded2cb600d5a91a196e76bf020

Test coverage of 'missing from store':

  $ hg serve -R ../mirror -d -p $HGPORT2 --pid-file hg.pid --config largefiles.usercache=$TESTTMP/mirrorcache --config "web.allow_push=*" --config web.push_ssl=no
  $ cat hg.pid >> $DAEMON_PIDS

  $ hg push http://localhost:$HGPORT2 -f --config largefiles.usercache=nocache
  pushing to http://localhost:$HGPORT2/
  searching for changes
  abort: largefile e2fb5f2139d086ded2cb600d5a91a196e76bf020 missing from store (needs to be uploaded)
  [255]

Verify that --lfrev controls which revisions are checked for largefiles to push

  $ hg push http://localhost:$HGPORT2 -f --config largefiles.usercache=nocache --lfrev tip
  pushing to http://localhost:$HGPORT2/
  searching for changes
  abort: largefile e2fb5f2139d086ded2cb600d5a91a196e76bf020 missing from store (needs to be uploaded)
  [255]

  $ hg push http://localhost:$HGPORT2 -f --config largefiles.usercache=nocache --lfrev null
  pushing to http://localhost:$HGPORT2/
  searching for changes
  remote: adding changesets
  remote: adding manifests
  remote: adding file changes
  remote: added 1 changesets with 1 changes to 1 files (+1 heads)

#endif