view tests/test-upgrade-repo.t @ 40936:3764330f76a6

sparse-revlog: enabled by default

The feature provides large benefits. It now seems mature enough to be enabled
by default.

* It solves catastrophic issues regarding delta storage in revlog,
* It allows for shorter delta chain in all repositories, improving
  performances.

Running benchmark of a wide range of operation did not reveal problematic
impact. Performance gains are observed where expected.

The format is supported by Mercurial version 4.7. So it seems safe to enable
it by default now.

Here is a reminder of key numbers regarding this delta strategy effect on
repository size and performance.

Effect on Size:
===============

For repositories with a lot of branches, sparse-revlog significantly improve
size, fixing limitation associated with the span of a delta chain. In
addition, sparse-revlog, deal well with limitations of the delta chain length.
For large repositories, this allows for a stiff reduction of the delta chain
without a problematic impact on the repository size. This delta chain length
improvement helps all repositories, not just the ones with many branches.


As a reminder, here are the default chain limits for each "format":
  * no-sparse: none
  * sparse:    1000

Mercurial
---------

    Manifest Size:
        limit       | none        |        1000
        ------------|-------------|------------
        no-sparse   |   6 143 044 |  6 269 496
        sparse      |   5 798 796 |  5 827 025

    Manifest Chain length data
        limit       ||       none        ||       1000
        value       || average |   max   || average |   max
        ------------||---------|---------||---------|---------
        no-sparse   ||     429 |   1 397 ||     397 |    1 000
        sparse      ||     326 |   1 290 ||     313 |    1 000

    Full Store Size
        limit       | none        |        1000
        ------------|-------------|------------
        no-sparse   |  46 944 775 | 47 166 129
        sparse      |  46 622 445 | 46 723 774

pypy
----

    Manifest Size:
        limit       | none        |        1000
        ------------|-------------|------------
        no-sparse   |  52 941 760 |  56 200 970
        sparse      |  26 348 229 |  27 384 133

    Manifest Chain length data
        limit       ||       none        ||       1000
        value       || average |   max   || average |   max
        ------------||---------|---------||---------|---------
        no-sparse   ||     769 |   3 889 ||     390 |    1 000
        sparse      ||   1 223 |   3 846 ||     495 |    1 000

    Full Store Size
        limit       | none        |        1000
        ------------|-------------|------------
        no-sparse   | 336 050 203 | 339 309 413
        sparse      | 338 673 985 | 339 709 889

Mozilla
-------

    Manifest Size:
        limit       | none           |        1000
        ------------|----------------|---------------
        no-sparse   |    215 096 339 |  1 708 853 525
        sparse      |    188 947 271 |    278 894 170

    Manifest Chain length data
        limit       ||       none        ||       1000
        value       || average |   max   || average |   max
        ------------||---------|---------||---------|--------
        no-sparse   ||  20 454 |  59 562 ||     491 |   1 000
        sparse      ||  23 509 |  69 891 ||     489 |   1 000

    Full Store Size
        limit       | none           |        1000
        ------------|----------------|---------------
        no-sparse   |  2 377 578 715 |  3 876 258 798
        sparse      |  2 441 677 137 |  2 535 997 381

Netbeans
--------

    Manifest Size:
        limit       | none           |        1000
        ------------|----------------|---------------
        no-sparse   |    130 088 982 |    741 590 565
        sparse      |    118 836 887 |    159 161 207

    Manifest Chain length data
        limit       ||       none        ||       1000
        value       || average |   max   || average |   max
        ------------||---------|---------||---------|---------
        no-sparse   ||  19 321 |  61 397 ||     510 |    1 000
        sparse      ||  21 240 |  61 583 ||     503 |    1 000

    Full Store Size
        limit       | none           |        1000
        ------------|----------------|---------------
        no-sparse   |  1 160 013 008 |  1 771 514 591
        sparse      |  1 164 959 988 |  1 205 284 308

Private repo #1
---------------

    Manifest Size:
        limit       | none            |        1000
        ------------|-----------------|---------------
        no-sparse   | 33 725 285 081  | 33 724 834 190
        sparse      |    350 542 420  |    423 470 579

     Manifest Chain length data
         limit       ||       none        ||       1000
         value       || average |   max   || average |   max
         ------------||---------|---------||---------|---------
         no-sparse   ||     282 |   8 885 ||     113 |    1 000
         snapshot    ||   3 655 |   8 951 ||     530 |    1 000

    Full Store Size
        limit       | none           |        1000
        ------------|----------------|---------------
        no-sparse   | 41 544 149 652 | 41 543 698 761
        sparse      |  8 448 037 300 |  8 520 965 459

Effect on speed:
================

Performances are strongly impacted by the delta chain length. Longer chain
results in slower revision restoration. For this reason, the 1000 chain limit
introduced by sparse-revlog helps repository with previously large chains a
lot. In our corpus, this means `netbeans` and `mozilla-central` who suffered
from unreasonable manifest delta chain length.

Another way sparse revlog helps, is by producing better delta's. For
repositories with many branches, the pathological patterns that resulted in
many sub-optimal deltas are gone. Smaller delta help with operations where
deltas are directly relevant, like bundle.

However, the sparse-revlog logic introduces some extra processing and a more
throughout testing of possible delta candidates. Adding an extra cost in some
cases. This cost is usually counterbalanced by the other performance gain.
However, for smaller repositories not affected by delta chain length issues or
branching related issues, this might make things a bit slower. However, these
are also repository where revlog performance is dwarfed by other costs.

Below are the summary of some timing from the performance test suite running
at `http://perf.octobus.net/` for a handful of key commands and operation.

It is important to keep in mind that most of this command works on the tip
part of the repository. The non-sparse and sparse version produce different
delta chains and the tip revision can end up at an arbitrary point of these
chains. This will impact some performance number listed in this summary.

For the record: here is the delta chain length for the tip revision of
manifest log in the benchmarked repository:

              |  no-sparse |   sparse   |
    mercurial |         94 |        904 |
    pypy      |         23 |        673 |
    netbeans  |       4158 |        258 |
    mozilla   |      63263 |        781 |

As you can see, the chain length for mercurial and pypy turn out to be
significantly longer. The netbeans and mozilla one get shorter because these
repositories benefit from the maximum chain length.

Timing for `hg commit`:
-----------------------

The time taken by `hg commit` does not varies significantly, no drawback for
using sparse here.

              |  no-sparse |   sparse   |
    mercurial |     68.1ms |     66.7ms |
    pypy      |     95.0ms |     94.1ms |
    netbeans  |    614.0ms |    611.0ms |
    mozilla   |   1340.0ms |  1.320.0ms |

Check the final section for statistics on a wider array of write.

Timing for bundling 10 000 changesets
-------------------------------------

The repository that benefits from better delta see a good performance boost.
The other ones are not significantly affected.

              |  no-sparse |   sparse  |
    mercurial |       3.1s |      3.0s |
    pypy      |      25.1s |      7.5s |
    netbeans  |      24.2s |     17.0s |
    mozilla   |      23.7s |     25.0s |

Timing for unbundling 1 000 changesets
--------------------------------------

Mercurial and mozilla are unaffected. The pypy repository benefit well from
the better delta.

However, the netbeans repository takes a visible hit. Digging that difference
reveals that it comes from the sparse-revlog bundle having to deal with a
snapshot that was re-encoded in the bundle. The slow path for adding new a revision
had to be triggered for it, slowing things down. The Sparse versions do not have
such snapshot to handle similar cases in the tested configuration.

              |  no-sparse |   sparse   |
    mercurial |      519ms |      502ms |
    pypy      |    1.270ms |      886ms |
    netbeans  |    1.370ms |    2.250ms |
    mozilla   |    3.230ms |    3.210ms |

Netbeans benefits from the better deltas in other dimensions too. For
example, the produced bundle is significantly smaller:

      * netbeans-no-sparse.hg: 2.3MB
      * netbeans-sparse.hg:    1.9MB


Timing to restore the tip most manifest entry:
----------------------------------------------

Nothing surprising here. The timing for mercurial and pypy are within a small
range where they won't affect performance much. In our tested case, they are
slower as they use a longer chain.

Timing for netbeans and mozilla improves a lot. Removing a significant amount
of time.

              |  no-sparse |   sparse   |
    mercurial |     1.09ms |     3.15ms |
    pypy      |     4.11ms |    10.70ms |
    netbeans  |   239.00ms |   112.00ms |
    mozilla   |   688.00ms |   198.00ms |

Reading 100 revision in descending order:
-----------------------------------------

We see the same kind of effect when reading the last 100 revisions. Large
boost for netbeans and mozilla, as they use much smaller delta chain.

Mercurial and pypy longer chain means slower reads, but nothing gets out of
control.

              |  no-sparse |   sparse   |
    mercurial |     0.089s |     0.268s |
    pypy      |     0.259s |     0.698s |
    netbeans  |   125.000s |    20.600s |
    mozilla   |    23.000s |    11.400s |


Writing from full text: statistic for the last 30K revisions
------------------------------------------------------------

This benchmark adds revisions to revlog from their full text. This is similar
to the work done during a commit, but for a large amount of revisions so that
we get a more relevant view.

We see better overall performances with sparse-revlog. The very worst case is
usually slower with sparse-revlog, but does not gets out of control. For the
vast majorities of the other writes, sparse-revlog is significantly faster for
larger repositories. This is reflected in the accumulated rewrite time for
netbeans and mozilla.

The notable exception is the pypy repository where things get slower. The
extra processing is not balanced by shorter delta chain. However, this is to
be seen as a blocking issue. First, the overall time spend dealing with revlog
for the repository pypy size is small compared to the other costs, so we get
slower on operations that matter less than for other larger repository.
Second, we still get nice size benefit from using sparse-revlog, smaller repo
size brings other usability and speed benefit (eg: bundle size).

max time  |  no-sparse |   sparse   |
mercurial |  0.010143s |  0.045280s |
pypy      |  0.034924s |  0.243288s |
netbeans  |  0.605371s |  2.130876s |
mozilla   |  1.478342s |  3.424541s |

99% time  |  no-sparse |   sparse   |
mercurial |  0.003774s |  0.003758s |
pypy      |  0.017387s |  0.025310s |
netbeans  |  0.576913s |  0.271195s |
mozilla   |  1.478342s |  0.449661s |

95% time  |  no-sparse |   sparse   |
mercurial |  0.002069s |  0.002120s |
pypy      |  0.010141s |  0.014797s |
netbeans  |  0.540202s |  0.258644s |
mozilla   |  0.654830s |  0.243440s |

full time |  no-sparse |   sparse  |
mercurial |     14.15s |    14.87s |
pypy      |     90.50s |   137.12s |
netbeans  |   6401.06s |  3411.14s |
mozilla   |   3086.89s |  1991.97s |

Differential Revision: https://phab.mercurial-scm.org/D5345

#require no-reposimplestore

  $ cat >> $HGRCPATH << EOF
  > [extensions]
  > share =
  > EOF

store and revlogv1 are required in source

  $ hg --config format.usestore=false init no-store
  $ hg -R no-store debugupgraderepo
  abort: cannot upgrade repository; requirement missing: store
  [255]

  $ hg init no-revlogv1
  $ cat > no-revlogv1/.hg/requires << EOF
  > dotencode
  > fncache
  > generaldelta
  > store
  > EOF

  $ hg -R no-revlogv1 debugupgraderepo
  abort: cannot upgrade repository; requirement missing: revlogv1
  [255]

Cannot upgrade shared repositories

  $ hg init share-parent
  $ hg -q share share-parent share-child

  $ hg -R share-child debugupgraderepo
  abort: cannot upgrade repository; unsupported source requirement: shared
  [255]

Do not yet support upgrading treemanifest repos

  $ hg --config experimental.treemanifest=true init treemanifest
  $ hg -R treemanifest debugupgraderepo
  abort: cannot upgrade repository; unsupported source requirement: treemanifest
  [255]

Cannot add treemanifest requirement during upgrade

  $ hg init disallowaddedreq
  $ hg -R disallowaddedreq --config experimental.treemanifest=true debugupgraderepo
  abort: cannot upgrade repository; do not support adding requirement: treemanifest
  [255]

An upgrade of a repository created with recommended settings only suggests optimizations

  $ hg init empty
  $ cd empty
  $ hg debugformat
  format-variant repo
  fncache:        yes
  dotencode:      yes
  generaldelta:   yes
  sparserevlog:   yes
  plain-cl-delta: yes
  compression:    zlib
  $ hg debugformat --verbose
  format-variant repo config default
  fncache:        yes    yes     yes
  dotencode:      yes    yes     yes
  generaldelta:   yes    yes     yes
  sparserevlog:   yes    yes     yes
  plain-cl-delta: yes    yes     yes
  compression:    zlib   zlib    zlib
  $ hg debugformat --verbose --config format.usefncache=no
  format-variant repo config default
  fncache:        yes     no     yes
  dotencode:      yes     no     yes
  generaldelta:   yes    yes     yes
  sparserevlog:   yes    yes     yes
  plain-cl-delta: yes    yes     yes
  compression:    zlib   zlib    zlib
  $ hg debugformat --verbose --config format.usefncache=no --color=debug
  format-variant repo config default
  [formatvariant.name.mismatchconfig|fncache:       ][formatvariant.repo.mismatchconfig| yes][formatvariant.config.special|     no][formatvariant.default|     yes]
  [formatvariant.name.mismatchconfig|dotencode:     ][formatvariant.repo.mismatchconfig| yes][formatvariant.config.special|     no][formatvariant.default|     yes]
  [formatvariant.name.uptodate|generaldelta:  ][formatvariant.repo.uptodate| yes][formatvariant.config.default|    yes][formatvariant.default|     yes]
  [formatvariant.name.uptodate|sparserevlog:  ][formatvariant.repo.uptodate| yes][formatvariant.config.default|    yes][formatvariant.default|     yes]
  [formatvariant.name.uptodate|plain-cl-delta:][formatvariant.repo.uptodate| yes][formatvariant.config.default|    yes][formatvariant.default|     yes]
  [formatvariant.name.uptodate|compression:   ][formatvariant.repo.uptodate| zlib][formatvariant.config.default|   zlib][formatvariant.default|    zlib]
  $ hg debugformat -Tjson
  [
   {
    "config": true,
    "default": true,
    "name": "fncache",
    "repo": true
   },
   {
    "config": true,
    "default": true,
    "name": "dotencode",
    "repo": true
   },
   {
    "config": true,
    "default": true,
    "name": "generaldelta",
    "repo": true
   },
   {
    "config": true,
    "default": true,
    "name": "sparserevlog",
    "repo": true
   },
   {
    "config": true,
    "default": true,
    "name": "plain-cl-delta",
    "repo": true
   },
   {
    "config": "zlib",
    "default": "zlib",
    "name": "compression",
    "repo": "zlib"
   }
  ]
  $ hg debugupgraderepo
  (no feature deficiencies found in existing repository)
  performing an upgrade with "--run" will make the following changes:
  
  requirements
     preserved: dotencode, fncache, generaldelta, revlogv1, sparserevlog, store
  
  additional optimizations are available by specifying "--optimize <name>":
  
  redeltaparent
     deltas within internal storage will be recalculated to choose an optimal base revision where this was not already done; the size of the repository may shrink and various operations may become faster; the first time this optimization is performed could slow down upgrade execution considerably; subsequent invocations should not run noticeably slower
  
  redeltamultibase
     deltas within internal storage will be recalculated against multiple base revision and the smallest difference will be used; the size of the repository may shrink significantly when there are many merges; this optimization will slow down execution in proportion to the number of merges in the repository and the amount of files in the repository; this slow down should not be significant unless there are tens of thousands of files and thousands of merges
  
  redeltaall
     deltas within internal storage will always be recalculated without reusing prior deltas; this will likely make execution run several times slower; this optimization is typically not needed
  
  redeltafulladd
     every revision will be re-added as if it was new content. It will go through the full storage mechanism giving extensions a chance to process it (eg. lfs). This is similar to "redeltaall" but even slower since more logic is involved.
  

--optimize can be used to add optimizations

  $ hg debugupgrade --optimize redeltaparent
  (no feature deficiencies found in existing repository)
  performing an upgrade with "--run" will make the following changes:
  
  requirements
     preserved: dotencode, fncache, generaldelta, revlogv1, sparserevlog, store
  
  redeltaparent
     deltas within internal storage will choose a new base revision if needed
  
  additional optimizations are available by specifying "--optimize <name>":
  
  redeltamultibase
     deltas within internal storage will be recalculated against multiple base revision and the smallest difference will be used; the size of the repository may shrink significantly when there are many merges; this optimization will slow down execution in proportion to the number of merges in the repository and the amount of files in the repository; this slow down should not be significant unless there are tens of thousands of files and thousands of merges
  
  redeltaall
     deltas within internal storage will always be recalculated without reusing prior deltas; this will likely make execution run several times slower; this optimization is typically not needed
  
  redeltafulladd
     every revision will be re-added as if it was new content. It will go through the full storage mechanism giving extensions a chance to process it (eg. lfs). This is similar to "redeltaall" but even slower since more logic is involved.
  

Various sub-optimal detections work

  $ cat > .hg/requires << EOF
  > revlogv1
  > store
  > EOF

  $ hg debugformat
  format-variant repo
  fncache:         no
  dotencode:       no
  generaldelta:    no
  sparserevlog:    no
  plain-cl-delta: yes
  compression:    zlib
  $ hg debugformat --verbose
  format-variant repo config default
  fncache:         no    yes     yes
  dotencode:       no    yes     yes
  generaldelta:    no    yes     yes
  sparserevlog:    no    yes     yes
  plain-cl-delta: yes    yes     yes
  compression:    zlib   zlib    zlib
  $ hg debugformat --verbose --config format.usegeneraldelta=no
  format-variant repo config default
  fncache:         no    yes     yes
  dotencode:       no    yes     yes
  generaldelta:    no     no     yes
  sparserevlog:    no     no     yes
  plain-cl-delta: yes    yes     yes
  compression:    zlib   zlib    zlib
  $ hg debugformat --verbose --config format.usegeneraldelta=no --color=debug
  format-variant repo config default
  [formatvariant.name.mismatchconfig|fncache:       ][formatvariant.repo.mismatchconfig|  no][formatvariant.config.default|    yes][formatvariant.default|     yes]
  [formatvariant.name.mismatchconfig|dotencode:     ][formatvariant.repo.mismatchconfig|  no][formatvariant.config.default|    yes][formatvariant.default|     yes]
  [formatvariant.name.mismatchdefault|generaldelta:  ][formatvariant.repo.mismatchdefault|  no][formatvariant.config.special|     no][formatvariant.default|     yes]
  [formatvariant.name.mismatchdefault|sparserevlog:  ][formatvariant.repo.mismatchdefault|  no][formatvariant.config.special|     no][formatvariant.default|     yes]
  [formatvariant.name.uptodate|plain-cl-delta:][formatvariant.repo.uptodate| yes][formatvariant.config.default|    yes][formatvariant.default|     yes]
  [formatvariant.name.uptodate|compression:   ][formatvariant.repo.uptodate| zlib][formatvariant.config.default|   zlib][formatvariant.default|    zlib]
  $ hg debugupgraderepo
  repository lacks features recommended by current config options:
  
  fncache
     long and reserved filenames may not work correctly; repository performance is sub-optimal
  
  dotencode
     storage of filenames beginning with a period or space may not work correctly
  
  generaldelta
     deltas within internal storage are unable to choose optimal revisions; repository is larger and slower than it could be; interaction with other repositories may require extra network and CPU resources, making "hg push" and "hg pull" slower
  
  sparserevlog
     in order to limit disk reading and memory usage on older version, the span of a delta chain from its root to its end is limited, whatever the relevant data in this span. This can severly limit Mercurial ability to build good chain of delta resulting is much more storage space being taken and limit reusability of on disk delta during exchange.
  
  
  performing an upgrade with "--run" will make the following changes:
  
  requirements
     preserved: revlogv1, store
     added: dotencode, fncache, generaldelta, sparserevlog
  
  fncache
     repository will be more resilient to storing certain paths and performance of certain operations should be improved
  
  dotencode
     repository will be better able to store files beginning with a space or period
  
  generaldelta
     repository storage will be able to create optimal deltas; new repository data will be smaller and read times should decrease; interacting with other repositories using this storage model should require less network and CPU resources, making "hg push" and "hg pull" faster
  
  sparserevlog
     Revlog supports delta chain with more unused data between payload. These gaps will be skipped at read time. This allows for better delta chains, making a better compression and faster exchange with server.
  
  additional optimizations are available by specifying "--optimize <name>":
  
  redeltaparent
     deltas within internal storage will be recalculated to choose an optimal base revision where this was not already done; the size of the repository may shrink and various operations may become faster; the first time this optimization is performed could slow down upgrade execution considerably; subsequent invocations should not run noticeably slower
  
  redeltamultibase
     deltas within internal storage will be recalculated against multiple base revision and the smallest difference will be used; the size of the repository may shrink significantly when there are many merges; this optimization will slow down execution in proportion to the number of merges in the repository and the amount of files in the repository; this slow down should not be significant unless there are tens of thousands of files and thousands of merges
  
  redeltaall
     deltas within internal storage will always be recalculated without reusing prior deltas; this will likely make execution run several times slower; this optimization is typically not needed
  
  redeltafulladd
     every revision will be re-added as if it was new content. It will go through the full storage mechanism giving extensions a chance to process it (eg. lfs). This is similar to "redeltaall" but even slower since more logic is involved.
  

  $ hg --config format.dotencode=false debugupgraderepo
  repository lacks features recommended by current config options:
  
  fncache
     long and reserved filenames may not work correctly; repository performance is sub-optimal
  
  generaldelta
     deltas within internal storage are unable to choose optimal revisions; repository is larger and slower than it could be; interaction with other repositories may require extra network and CPU resources, making "hg push" and "hg pull" slower
  
  sparserevlog
     in order to limit disk reading and memory usage on older version, the span of a delta chain from its root to its end is limited, whatever the relevant data in this span. This can severly limit Mercurial ability to build good chain of delta resulting is much more storage space being taken and limit reusability of on disk delta during exchange.
  
  repository lacks features used by the default config options:
  
  dotencode
     storage of filenames beginning with a period or space may not work correctly
  
  
  performing an upgrade with "--run" will make the following changes:
  
  requirements
     preserved: revlogv1, store
     added: fncache, generaldelta, sparserevlog
  
  fncache
     repository will be more resilient to storing certain paths and performance of certain operations should be improved
  
  generaldelta
     repository storage will be able to create optimal deltas; new repository data will be smaller and read times should decrease; interacting with other repositories using this storage model should require less network and CPU resources, making "hg push" and "hg pull" faster
  
  sparserevlog
     Revlog supports delta chain with more unused data between payload. These gaps will be skipped at read time. This allows for better delta chains, making a better compression and faster exchange with server.
  
  additional optimizations are available by specifying "--optimize <name>":
  
  redeltaparent
     deltas within internal storage will be recalculated to choose an optimal base revision where this was not already done; the size of the repository may shrink and various operations may become faster; the first time this optimization is performed could slow down upgrade execution considerably; subsequent invocations should not run noticeably slower
  
  redeltamultibase
     deltas within internal storage will be recalculated against multiple base revision and the smallest difference will be used; the size of the repository may shrink significantly when there are many merges; this optimization will slow down execution in proportion to the number of merges in the repository and the amount of files in the repository; this slow down should not be significant unless there are tens of thousands of files and thousands of merges
  
  redeltaall
     deltas within internal storage will always be recalculated without reusing prior deltas; this will likely make execution run several times slower; this optimization is typically not needed
  
  redeltafulladd
     every revision will be re-added as if it was new content. It will go through the full storage mechanism giving extensions a chance to process it (eg. lfs). This is similar to "redeltaall" but even slower since more logic is involved.
  

  $ cd ..

Upgrading a repository that is already modern essentially no-ops

  $ hg init modern
  $ hg -R modern debugupgraderepo --run
  upgrade will perform the following actions:
  
  requirements
     preserved: dotencode, fncache, generaldelta, revlogv1, sparserevlog, store
  
  beginning upgrade...
  repository locked and read-only
  creating temporary repository to stage migrated data: $TESTTMP/modern/.hg/upgrade.* (glob)
  (it is safe to interrupt this process any time before data migration completes)
  data fully migrated to temporary repository
  marking source repository as being upgraded; clients will be unable to read from repository
  starting in-place swap of repository data
  replaced files will be backed up at $TESTTMP/modern/.hg/upgradebackup.* (glob)
  replacing store...
  store replacement complete; repository was inconsistent for *s (glob)
  finalizing requirements file and making repository readable again
  removing temporary repository $TESTTMP/modern/.hg/upgrade.* (glob)
  copy of old repository backed up at $TESTTMP/modern/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified

Upgrading a repository to generaldelta works

  $ hg --config format.usegeneraldelta=false init upgradegd
  $ cd upgradegd
  $ touch f0
  $ hg -q commit -A -m initial
  $ touch f1
  $ hg -q commit -A -m 'add f1'
  $ hg -q up -r 0
  $ touch f2
  $ hg -q commit -A -m 'add f2'

  $ hg debugupgraderepo --run --config format.sparse-revlog=false
  upgrade will perform the following actions:
  
  requirements
     preserved: dotencode, fncache, revlogv1, store
     added: generaldelta
  
  generaldelta
     repository storage will be able to create optimal deltas; new repository data will be smaller and read times should decrease; interacting with other repositories using this storage model should require less network and CPU resources, making "hg push" and "hg pull" faster
  
  beginning upgrade...
  repository locked and read-only
  creating temporary repository to stage migrated data: $TESTTMP/upgradegd/.hg/upgrade.* (glob)
  (it is safe to interrupt this process any time before data migration completes)
  migrating 9 total revisions (3 in filelogs, 3 in manifests, 3 in changelog)
  migrating 917 bytes in store; 401 bytes tracked data
  migrating 3 filelogs containing 3 revisions (192 bytes in store; 0 bytes tracked data)
  finished migrating 3 filelog revisions across 3 filelogs; change in size: 0 bytes
  migrating 1 manifests containing 3 revisions (349 bytes in store; 220 bytes tracked data)
  finished migrating 3 manifest revisions across 1 manifests; change in size: 0 bytes
  migrating changelog containing 3 revisions (376 bytes in store; 181 bytes tracked data)
  finished migrating 3 changelog revisions; change in size: 0 bytes
  finished migrating 9 total revisions; total change in store size: 0 bytes
  copying phaseroots
  data fully migrated to temporary repository
  marking source repository as being upgraded; clients will be unable to read from repository
  starting in-place swap of repository data
  replaced files will be backed up at $TESTTMP/upgradegd/.hg/upgradebackup.* (glob)
  replacing store...
  store replacement complete; repository was inconsistent for *s (glob)
  finalizing requirements file and making repository readable again
  removing temporary repository $TESTTMP/upgradegd/.hg/upgrade.* (glob)
  copy of old repository backed up at $TESTTMP/upgradegd/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified

Original requirements backed up

  $ cat .hg/upgradebackup.*/requires
  dotencode
  fncache
  revlogv1
  store

generaldelta added to original requirements files

  $ cat .hg/requires
  dotencode
  fncache
  generaldelta
  revlogv1
  store

store directory has files we expect

  $ ls .hg/store
  00changelog.i
  00manifest.i
  data
  fncache
  phaseroots
  undo
  undo.backupfiles
  undo.phaseroots

manifest should be generaldelta

  $ hg debugrevlog -m | grep flags
  flags  : inline, generaldelta

verify should be happy

  $ hg verify
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  checked 3 changesets with 3 changes to 3 files

old store should be backed up

  $ ls .hg/upgradebackup.*/store
  00changelog.i
  00manifest.i
  data
  fncache
  phaseroots
  undo
  undo.backup.fncache
  undo.backupfiles
  undo.phaseroots

  $ cd ..

store files with special filenames aren't encoded during copy

  $ hg init store-filenames
  $ cd store-filenames
  $ touch foo
  $ hg -q commit -A -m initial
  $ touch .hg/store/.XX_special_filename

  $ hg debugupgraderepo --run
  upgrade will perform the following actions:
  
  requirements
     preserved: dotencode, fncache, generaldelta, revlogv1, sparserevlog, store
  
  beginning upgrade...
  repository locked and read-only
  creating temporary repository to stage migrated data: $TESTTMP/store-filenames/.hg/upgrade.* (glob)
  (it is safe to interrupt this process any time before data migration completes)
  migrating 3 total revisions (1 in filelogs, 1 in manifests, 1 in changelog)
  migrating 301 bytes in store; 107 bytes tracked data
  migrating 1 filelogs containing 1 revisions (64 bytes in store; 0 bytes tracked data)
  finished migrating 1 filelog revisions across 1 filelogs; change in size: 0 bytes
  migrating 1 manifests containing 1 revisions (110 bytes in store; 45 bytes tracked data)
  finished migrating 1 manifest revisions across 1 manifests; change in size: 0 bytes
  migrating changelog containing 1 revisions (127 bytes in store; 62 bytes tracked data)
  finished migrating 1 changelog revisions; change in size: 0 bytes
  finished migrating 3 total revisions; total change in store size: 0 bytes
  copying .XX_special_filename
  copying phaseroots
  data fully migrated to temporary repository
  marking source repository as being upgraded; clients will be unable to read from repository
  starting in-place swap of repository data
  replaced files will be backed up at $TESTTMP/store-filenames/.hg/upgradebackup.* (glob)
  replacing store...
  store replacement complete; repository was inconsistent for *s (glob)
  finalizing requirements file and making repository readable again
  removing temporary repository $TESTTMP/store-filenames/.hg/upgrade.* (glob)
  copy of old repository backed up at $TESTTMP/store-filenames/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified
  $ hg debugupgraderepo --run --optimize redeltafulladd
  upgrade will perform the following actions:
  
  requirements
     preserved: dotencode, fncache, generaldelta, revlogv1, sparserevlog, store
  
  redeltafulladd
     each revision will be added as new content to the internal storage; this will likely drastically slow down execution time, but some extensions might need it
  
  beginning upgrade...
  repository locked and read-only
  creating temporary repository to stage migrated data: $TESTTMP/store-filenames/.hg/upgrade.* (glob)
  (it is safe to interrupt this process any time before data migration completes)
  migrating 3 total revisions (1 in filelogs, 1 in manifests, 1 in changelog)
  migrating 301 bytes in store; 107 bytes tracked data
  migrating 1 filelogs containing 1 revisions (64 bytes in store; 0 bytes tracked data)
  finished migrating 1 filelog revisions across 1 filelogs; change in size: 0 bytes
  migrating 1 manifests containing 1 revisions (110 bytes in store; 45 bytes tracked data)
  finished migrating 1 manifest revisions across 1 manifests; change in size: 0 bytes
  migrating changelog containing 1 revisions (127 bytes in store; 62 bytes tracked data)
  finished migrating 1 changelog revisions; change in size: 0 bytes
  finished migrating 3 total revisions; total change in store size: 0 bytes
  copying .XX_special_filename
  copying phaseroots
  data fully migrated to temporary repository
  marking source repository as being upgraded; clients will be unable to read from repository
  starting in-place swap of repository data
  replaced files will be backed up at $TESTTMP/store-filenames/.hg/upgradebackup.* (glob)
  replacing store...
  store replacement complete; repository was inconsistent for *s (glob)
  finalizing requirements file and making repository readable again
  removing temporary repository $TESTTMP/store-filenames/.hg/upgrade.* (glob)
  copy of old repository backed up at $TESTTMP/store-filenames/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified

fncache is valid after upgrade

  $ hg debugrebuildfncache
  fncache already up to date

  $ cd ..

Check upgrading a large file repository
---------------------------------------

  $ hg init largefilesrepo
  $ cat << EOF >> largefilesrepo/.hg/hgrc
  > [extensions]
  > largefiles =
  > EOF

  $ cd largefilesrepo
  $ touch foo
  $ hg add --large foo
  $ hg -q commit -m initial
  $ cat .hg/requires
  dotencode
  fncache
  generaldelta
  largefiles
  revlogv1
  sparserevlog
  store

  $ hg debugupgraderepo --run
  upgrade will perform the following actions:
  
  requirements
     preserved: dotencode, fncache, generaldelta, largefiles, revlogv1, sparserevlog, store
  
  beginning upgrade...
  repository locked and read-only
  creating temporary repository to stage migrated data: $TESTTMP/largefilesrepo/.hg/upgrade.* (glob)
  (it is safe to interrupt this process any time before data migration completes)
  migrating 3 total revisions (1 in filelogs, 1 in manifests, 1 in changelog)
  migrating 355 bytes in store; 160 bytes tracked data
  migrating 1 filelogs containing 1 revisions (106 bytes in store; 41 bytes tracked data)
  finished migrating 1 filelog revisions across 1 filelogs; change in size: 0 bytes
  migrating 1 manifests containing 1 revisions (116 bytes in store; 51 bytes tracked data)
  finished migrating 1 manifest revisions across 1 manifests; change in size: 0 bytes
  migrating changelog containing 1 revisions (133 bytes in store; 68 bytes tracked data)
  finished migrating 1 changelog revisions; change in size: 0 bytes
  finished migrating 3 total revisions; total change in store size: 0 bytes
  copying phaseroots
  data fully migrated to temporary repository
  marking source repository as being upgraded; clients will be unable to read from repository
  starting in-place swap of repository data
  replaced files will be backed up at $TESTTMP/largefilesrepo/.hg/upgradebackup.* (glob)
  replacing store...
  store replacement complete; repository was inconsistent for *s (glob)
  finalizing requirements file and making repository readable again
  removing temporary repository $TESTTMP/largefilesrepo/.hg/upgrade.* (glob)
  copy of old repository backed up at $TESTTMP/largefilesrepo/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified
  $ cat .hg/requires
  dotencode
  fncache
  generaldelta
  largefiles
  revlogv1
  sparserevlog
  store

  $ cat << EOF >> .hg/hgrc
  > [extensions]
  > lfs =
  > [lfs]
  > threshold = 10
  > EOF
  $ echo '123456789012345' > lfs.bin
  $ hg ci -Am 'lfs.bin'
  adding lfs.bin
  $ grep lfs .hg/requires
  lfs
  $ find .hg/store/lfs -type f
  .hg/store/lfs/objects/d0/beab232adff5ba365880366ad30b1edb85c4c5372442b5d2fe27adc96d653f

  $ hg debugupgraderepo --run
  upgrade will perform the following actions:
  
  requirements
     preserved: dotencode, fncache, generaldelta, largefiles, lfs, revlogv1, sparserevlog, store
  
  beginning upgrade...
  repository locked and read-only
  creating temporary repository to stage migrated data: $TESTTMP/largefilesrepo/.hg/upgrade.* (glob)
  (it is safe to interrupt this process any time before data migration completes)
  migrating 6 total revisions (2 in filelogs, 2 in manifests, 2 in changelog)
  migrating 801 bytes in store; 467 bytes tracked data
  migrating 2 filelogs containing 2 revisions (296 bytes in store; 182 bytes tracked data)
  finished migrating 2 filelog revisions across 2 filelogs; change in size: 0 bytes
  migrating 1 manifests containing 2 revisions (241 bytes in store; 151 bytes tracked data)
  finished migrating 2 manifest revisions across 1 manifests; change in size: 0 bytes
  migrating changelog containing 2 revisions (264 bytes in store; 134 bytes tracked data)
  finished migrating 2 changelog revisions; change in size: 0 bytes
  finished migrating 6 total revisions; total change in store size: 0 bytes
  copying phaseroots
  copying lfs blob d0beab232adff5ba365880366ad30b1edb85c4c5372442b5d2fe27adc96d653f
  data fully migrated to temporary repository
  marking source repository as being upgraded; clients will be unable to read from repository
  starting in-place swap of repository data
  replaced files will be backed up at $TESTTMP/largefilesrepo/.hg/upgradebackup.* (glob)
  replacing store...
  store replacement complete; repository was inconsistent for *s (glob)
  finalizing requirements file and making repository readable again
  removing temporary repository $TESTTMP/largefilesrepo/.hg/upgrade.* (glob)
  copy of old repository backed up at $TESTTMP/largefilesrepo/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified

  $ grep lfs .hg/requires
  lfs
  $ find .hg/store/lfs -type f
  .hg/store/lfs/objects/d0/beab232adff5ba365880366ad30b1edb85c4c5372442b5d2fe27adc96d653f
  $ hg verify
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  checked 2 changesets with 2 changes to 2 files
  $ hg debugdata lfs.bin 0
  version https://git-lfs.github.com/spec/v1
  oid sha256:d0beab232adff5ba365880366ad30b1edb85c4c5372442b5d2fe27adc96d653f
  size 16
  x-is-binary 0

  $ cd ..

repository config is taken in account
-------------------------------------

  $ cat << EOF >> $HGRCPATH
  > [format]
  > maxchainlen = 1
  > EOF

  $ hg init localconfig
  $ cd localconfig
  $ cat << EOF > file
  > some content
  > with some length
  > to make sure we get a delta
  > after changes
  > very long
  > very long
  > very long
  > very long
  > very long
  > very long
  > very long
  > very long
  > very long
  > very long
  > very long
  > EOF
  $ hg -q commit -A -m A
  $ echo "new line" >> file
  $ hg -q commit -m B
  $ echo "new line" >> file
  $ hg -q commit -m C

  $ cat << EOF >> .hg/hgrc
  > [format]
  > maxchainlen = 9001
  > EOF
  $ hg config format
  format.maxchainlen=9001
  $ hg debugdeltachain file
      rev  chain# chainlen     prev   delta       size    rawsize  chainsize     ratio   lindist extradist extraratio   readsize largestblk rddensity srchunks
        0       1        1       -1    base         77        182         77   0.42308        77         0    0.00000         77         77   1.00000        1
        1       1        2        0      p1         21        191         98   0.51309        98         0    0.00000         98         98   1.00000        1
        2       1        2        0   other         30        200        107   0.53500       128        21    0.19626        128        128   0.83594        1

  $ hg debugupgraderepo --run --optimize redeltaall
  upgrade will perform the following actions:
  
  requirements
     preserved: dotencode, fncache, generaldelta, revlogv1, sparserevlog, store
  
  redeltaall
     deltas within internal storage will be fully recomputed; this will likely drastically slow down execution time
  
  beginning upgrade...
  repository locked and read-only
  creating temporary repository to stage migrated data: $TESTTMP/localconfig/.hg/upgrade.* (glob)
  (it is safe to interrupt this process any time before data migration completes)
  migrating 9 total revisions (3 in filelogs, 3 in manifests, 3 in changelog)
  migrating 1019 bytes in store; 882 bytes tracked data
  migrating 1 filelogs containing 3 revisions (320 bytes in store; 573 bytes tracked data)
  finished migrating 3 filelog revisions across 1 filelogs; change in size: -9 bytes
  migrating 1 manifests containing 3 revisions (333 bytes in store; 138 bytes tracked data)
  finished migrating 3 manifest revisions across 1 manifests; change in size: 0 bytes
  migrating changelog containing 3 revisions (366 bytes in store; 171 bytes tracked data)
  finished migrating 3 changelog revisions; change in size: 0 bytes
  finished migrating 9 total revisions; total change in store size: -9 bytes
  copying phaseroots
  data fully migrated to temporary repository
  marking source repository as being upgraded; clients will be unable to read from repository
  starting in-place swap of repository data
  replaced files will be backed up at $TESTTMP/localconfig/.hg/upgradebackup.* (glob)
  replacing store...
  store replacement complete; repository was inconsistent for *s (glob)
  finalizing requirements file and making repository readable again
  removing temporary repository $TESTTMP/localconfig/.hg/upgrade.* (glob)
  copy of old repository backed up at $TESTTMP/localconfig/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified
  $ hg debugdeltachain file
      rev  chain# chainlen     prev   delta       size    rawsize  chainsize     ratio   lindist extradist extraratio   readsize largestblk rddensity srchunks
        0       1        1       -1    base         77        182         77   0.42308        77         0    0.00000         77         77   1.00000        1
        1       1        2        0      p1         21        191         98   0.51309        98         0    0.00000         98         98   1.00000        1
        2       1        3        1      p1         21        200        119   0.59500       119         0    0.00000        119        119   1.00000        1
  $ cd ..

  $ cat << EOF >> $HGRCPATH
  > [format]
  > maxchainlen = 9001
  > EOF

Check upgrading a sparse-revlog repository
---------------------------------------

  $ hg init sparserevlogrepo --config format.sparse-revlog=no
  $ cd sparserevlogrepo
  $ touch foo
  $ hg add foo
  $ hg -q commit -m "foo"
  $ cat .hg/requires
  dotencode
  fncache
  generaldelta
  revlogv1
  store

Check that we can add the sparse-revlog format requirement
  $ hg --config format.sparse-revlog=yes debugupgraderepo --run >/dev/null
  copy of old repository backed up at $TESTTMP/sparserevlogrepo/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified
  $ cat .hg/requires
  dotencode
  fncache
  generaldelta
  revlogv1
  sparserevlog
  store

Check that we can remove the sparse-revlog format requirement
  $ hg --config format.sparse-revlog=no debugupgraderepo --run >/dev/null
  copy of old repository backed up at $TESTTMP/sparserevlogrepo/.hg/upgradebackup.* (glob)
  the old repository will not be deleted; remove it to free up disk space once the upgraded repository is verified
  $ cat .hg/requires
  dotencode
  fncache
  generaldelta
  revlogv1
  store
  $ cd ..