tests/test-remotefilelog-gcrepack.t

match: convert O(n) to O(log n) in exactmatcher.visitchildrenset

When using narrow, during rebase this is called (at least) once per directory in
the set of files in the commit being rebased. Every time it's called, we did the
set arithmetic (now extracted and cached), which was probably pretty cheap but
not necessary to repeat each time, looped over every item in the matcher and
kept things that started with the directory we were querying.

With very large narrowspecs, and a commit that touched a file in a large number
of directories, this was slow. In a pathological repo, the rebase of a single
commit (that touched over 17k files, I believe in approximately as many
directories) with a narrowspec that had >32k entries took 8,246s of profiled
time, with 5,007s of that spent in visitchildrenset (transitively). With this
change, the time spent in visitchildrenset is less than 34s (which is where my
profile cut off). Most of the remaining time was network access due to our
custom remotefilelog-based setup not properly prefetching.

Differential Revision: https://phab.mercurial-scm.org/D10294

#require no-windows

  $ . "$TESTDIR/remotefilelog-library.sh"

  $ hg init master
  $ cd master
  $ cat >> .hg/hgrc <<EOF
  > [remotefilelog]
  > server=True
  > EOF
  $ echo x > x
  $ hg commit -qAm x
  $ echo y > y
  $ rm x
  $ hg commit -qAm DxAy
  $ echo yy > y
  $ hg commit -qAm y
  $ cd ..

  $ hgcloneshallow ssh://user@dummy/master shallow -q
  1 files fetched over 1 fetches - (1 misses, 0.00% hit ratio) over *s (glob)

# Set the prefetchdays config to zero so that all commits are prefetched
# no matter what their creation date is.
  $ cd shallow
  $ cat >> .hg/hgrc <<EOF
  > [remotefilelog]
  > prefetchdays=0
  > EOF
  $ cd ..

# Prefetch all data and repack

  $ cd shallow
  $ cat >> .hg/hgrc <<EOF
  > [remotefilelog]
  > bgprefetchrevs=all()
  > EOF

  $ hg prefetch
  2 files fetched over 1 fetches - (2 misses, 0.00% hit ratio) over *s (glob)
  $ hg repack

  $ find $CACHEDIR | sort | egrep ".datapack|.histpack"
  $TESTTMP/hgcache/master/packs/7bcd2d90b99395ca43172a0dd24e18860b2902f9.histpack
  $TESTTMP/hgcache/master/packs/dc8f8fdc76690ce27791ce9f53a18da379e50d37.datapack

# Ensure that all file versions were prefetched

  $ hg debugdatapack `ls -ct $TESTTMP/hgcache/master/packs/*.datapack | head -n 1`
  $TESTTMP/hgcache/master/packs/dc8f8fdc76690ce27791ce9f53a18da379e50d37:
  x:
  Node          Delta Base    Delta Length  Blob Size
  1406e7411862  000000000000  2             2
  
  Total:                      2             2         (0.0% bigger)
  y:
  Node          Delta Base    Delta Length  Blob Size
  50dbc4572b8e  000000000000  3             3
  076f5e2225b3  50dbc4572b8e  14            2
  
  Total:                      17            5         (240.0% bigger)

# Test garbage collection during repack

  $ cat >> .hg/hgrc <<EOF
  > [remotefilelog]
  > bgprefetchrevs=tip
  > gcrepack=True
  > nodettl=86400
  > EOF

  $ hg repack

  $ find $CACHEDIR | sort | egrep ".datapack|.histpack"
  $TESTTMP/hgcache/master/packs/7bcd2d90b99395ca43172a0dd24e18860b2902f9.histpack
  $TESTTMP/hgcache/master/packs/a4e1d094ec2aee8a08a4d6d95a13c634cc7d7394.datapack

# Ensure that file 'x' was garbage collected. It should be GCed because it is not in the keepset
# and is old (commit date is 0.0 in tests). Ensure that file 'y' is present as it is in the keepset.

  $ hg debugdatapack `ls -ct $TESTTMP/hgcache/master/packs/*.datapack | head -n 1`
  $TESTTMP/hgcache/master/packs/a4e1d094ec2aee8a08a4d6d95a13c634cc7d7394:
  y:
  Node          Delta Base    Delta Length  Blob Size
  50dbc4572b8e  000000000000  3             3
  
  Total:                      3             3         (0.0% bigger)

# Prefetch all data again and repack for later garbage collection

  $ cat >> .hg/hgrc <<EOF
  > [remotefilelog]
  > bgprefetchrevs=all()
  > EOF

  $ hg prefetch
  2 files fetched over 1 fetches - (2 misses, 0.00% hit ratio) over *s (glob)
  $ hg repack

  $ find $CACHEDIR | sort | egrep ".datapack|.histpack"
  $TESTTMP/hgcache/master/packs/7bcd2d90b99395ca43172a0dd24e18860b2902f9.histpack
  $TESTTMP/hgcache/master/packs/dc8f8fdc76690ce27791ce9f53a18da379e50d37.datapack

# Ensure that all file versions were prefetched

  $ hg debugdatapack `ls -ct $TESTTMP/hgcache/master/packs/*.datapack | head -n 1`
  $TESTTMP/hgcache/master/packs/dc8f8fdc76690ce27791ce9f53a18da379e50d37:
  x:
  Node          Delta Base    Delta Length  Blob Size
  1406e7411862  000000000000  2             2
  
  Total:                      2             2         (0.0% bigger)
  y:
  Node          Delta Base    Delta Length  Blob Size
  50dbc4572b8e  000000000000  3             3
  076f5e2225b3  50dbc4572b8e  14            2
  
  Total:                      17            5         (240.0% bigger)

# Test garbage collection during repack. Ensure that new files are not removed even though they are not in the keepset
# For the purposes of the test the TTL of a file is set to current time + 100 seconds. i.e. all commits in tests have
# a date of 1970 and therefore to prevent garbage collection we have to set nodettl to be farther from 1970 than we are now.

  $ cat >> .hg/hgrc <<EOF
  > [remotefilelog]
  > bgprefetchrevs=
  > nodettl=$(($(date +%s) + 100))
  > EOF

  $ hg repack

  $ find $CACHEDIR | sort | egrep ".datapack|.histpack"
  $TESTTMP/hgcache/master/packs/7bcd2d90b99395ca43172a0dd24e18860b2902f9.histpack
  $TESTTMP/hgcache/master/packs/dc8f8fdc76690ce27791ce9f53a18da379e50d37.datapack

# Ensure that all file versions were prefetched

  $ hg debugdatapack `ls -ct $TESTTMP/hgcache/master/packs/*.datapack | head -n 1`
  $TESTTMP/hgcache/master/packs/dc8f8fdc76690ce27791ce9f53a18da379e50d37:
  x:
  Node          Delta Base    Delta Length  Blob Size
  1406e7411862  000000000000  2             2
  
  Total:                      2             2         (0.0% bigger)
  y:
  Node          Delta Base    Delta Length  Blob Size
  50dbc4572b8e  000000000000  3             3
  076f5e2225b3  50dbc4572b8e  14            2
  
  Total:                      17            5         (240.0% bigger)