view tests/test-obsolete-checkheads.t @ 47010:76ae43d5b1db stable

repoview: fix memory leak of filtered repo classes

The leak occurs in long-running server processes with
extensions, and is measured at 110kB per request.

Before this change, the contents of the `_filteredrepotypes`
cache are not properly garbage collected, despite it begin
a `WeakKeyDictionary`.

Extensions have a tendency to generate a new repository class
for each `localrepo` instantiation. Server processes based
on `hgwebdir_mod` will instantiate a new `localrepo` for each
HTTP request that involves a repository.

As a result, with a testing process that repeatedly opens a
repository with several extensions activated
(`topic` notably among them), we see a steady increase in
resident memory of 110kB per repository instantiation before this
change. This is also true, if we call `gc.collect()` at each
instantiation, like `hgwebdir_mod` does, or not.

The cause of the leak is that the *values* aren't weak references.
This change uses `weakref.ref` for the values, and this makes
in our measurements the resident size increase drop to 5kB per
repository instantiation, with no explicit call of `gc.collect()`
at all.
There is currently no reason to believe that this remaining leak
of 5kB is related to or even due to Mercurial core.

We've also seen evidence that `ui.ui` instances weren't properly
garbage collected before the change (with the change, they are).
This could explain why the figures are relatively high.

In theory, the collection of weak references could lead to
much more misses in the cache, so we measured the impact on
the original case that was motivation for introducing that cache
in 7e89bd0cfb86 (see also issue5043): `hg convert` of the
mozilla-central repository. The bad news here is that there is a
major memory leak there, both with and without the present changeset.
There were no more cache misses, and we could see no
more memory leak with this change: the resident size after importing
roughly 100000 changesets was at 12.4GB before, and 12.5GB after.
The small increase is mentioned for completeness only, and we
believe that it should be ignored, at least as long as the main
leak isn't fixed. At less than 1% of the main leak, even finding out
whether it is merely noise would be wasteful.

Original context where this was spotted and first mitigated:
  https://foss.heptapod.net/heptapod/heptapod/-/issues/466

The leak reduction was also obtained in Heptapod inner HTTP server,
which amounts to the same as `hgwebdir_mod` for these questions.

The measurements done with Python 3.9, similar figures seen with 3.8.
More work on our side would be needed to give measurements with 2.7,
because of testing server process does not support it.

Check that obsolete properly strip heads
  $ cat >> $HGRCPATH << EOF
  > [phases]
  > # public changeset are not obsolete
  > publish=false
  > [command-templates]
  > log='{node|short} ({phase}) {desc|firstline}\n'
  > [experimental]
  > evolution.createmarkers=True
  > EOF
  $ mkcommit() {
  >    echo "$1" > "$1"
  >    hg add "$1"
  >    hg ci -m "add $1"
  > }
  $ getid() {
  >    hg id --debug -ir "desc('$1')"
  > }


  $ hg init remote
  $ cd remote
  $ mkcommit base
  $ hg phase --public .
  $ cd ..
  $ cp -R remote base
  $ hg clone remote local
  updating to branch default
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ cd local

New head replaces old head
==========================

setup
(we add the 1 flags to prevent bumped error during the test)

  $ mkcommit old
  $ hg push
  pushing to $TESTTMP/remote
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  $ hg up -q '.^'
  $ mkcommit new
  created new head
  $ hg debugobsolete --flags 1 `getid old` `getid new`
  1 new obsolescence markers
  obsoleted 1 changesets
  $ hg log -G --hidden
  @  71e3228bffe1 (draft) add new
  |
  | x  c70b08862e08 (draft) add old
  |/
  o  b4952fcf48cf (public) add base
  
  $ cp -R ../remote ../backup1

old exists remotely as draft. It is obsoleted by new that we now push.
Push should not warn about creating new head

  $ hg push
  pushing to $TESTTMP/remote
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)

old head is now public (public local version)
=============================================

setup

  $ rm -fr ../remote
  $ cp -R ../backup1 ../remote
  $ hg -R ../remote phase --public c70b08862e08
  $ hg pull -v
  pulling from $TESTTMP/remote
  searching for changes
  no changes found
  1 local changesets published
  $ hg log -G --hidden
  @  71e3228bffe1 (draft) add new
  |
  | o  c70b08862e08 (public) add old
  |/
  o  b4952fcf48cf (public) add base
  

Abort: old will still be an head because it's public.

  $ hg push
  pushing to $TESTTMP/remote
  searching for changes
  abort: push creates new remote head 71e3228bffe1
  (merge or see 'hg help push' for details about pushing new heads)
  [20]

old head is now public (public remote version)
==============================================

TODO: Not implemented yet.

# setup
#
#   $ rm -fr ../remote
#   $ cp -R ../backup1 ../remote
#   $ hg -R ../remote phase --public c70b08862e08
#   $ hg phase --draft --force c70b08862e08
#   $ hg log -G --hidden
#   @  71e3228bffe1 (draft) add new
#   |
#   | x  c70b08862e08 (draft) add old
#   |/
#   o  b4952fcf48cf (public) add base
#
#
#
# Abort: old will still be an head because it's public.
#
#   $ hg push
#   pushing to $TESTTMP/remote
#   searching for changes
#   abort: push creates new remote head 71e3228bffe1!
#   (merge or see 'hg help push' for details about pushing new heads)
#   [255]

old head is obsolete but replacement is not pushed
==================================================

setup

  $ rm -fr ../remote
  $ cp -R ../backup1 ../remote
  $ hg phase --draft --force '(0::) - 0'
  $ hg up -q '.^'
  $ mkcommit other
  created new head
  $ hg log -G --hidden
  @  d7d41ccbd4de (draft) add other
  |
  | o  71e3228bffe1 (draft) add new
  |/
  | x  c70b08862e08 (draft) add old
  |/
  o  b4952fcf48cf (public) add base
  

old exists remotely as draft. It is obsoleted by new but we don't push new.
Push should abort on new head

  $ hg push -r 'desc("other")'
  pushing to $TESTTMP/remote
  searching for changes
  abort: push creates new remote head d7d41ccbd4de
  (merge or see 'hg help push' for details about pushing new heads)
  [20]



Both precursors and successors are already know remotely. Descendant adds heads
===============================================================================

setup. (The obsolete marker is known locally only

  $ cd ..
  $ rm -rf local
  $ hg clone remote local
  updating to branch default
  2 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ cd local
  $ mkcommit old
  old already tracked!
  nothing changed
  [1]
  $ hg up -q '.^'
  $ mkcommit new
  created new head
  $ hg push -f
  pushing to $TESTTMP/remote
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)
  $ mkcommit desc1
  $ hg up -q '.^'
  $ mkcommit desc2
  created new head
  $ hg debugobsolete `getid old` `getid new`
  1 new obsolescence markers
  obsoleted 1 changesets
  $ hg log -G --hidden
  @  5fe37041cc2b (draft) add desc2
  |
  | o  a3ef1d111c5f (draft) add desc1
  |/
  o  71e3228bffe1 (draft) add new
  |
  | x  c70b08862e08 (draft) add old
  |/
  o  b4952fcf48cf (public) add base
  
  $ hg log -G --hidden -R ../remote
  o  71e3228bffe1 (draft) add new
  |
  | o  c70b08862e08 (draft) add old
  |/
  @  b4952fcf48cf (public) add base
  
  $ cp -R ../remote ../backup2

Push should not warn about adding new heads. We create one, but we'll delete
one anyway.

  $ hg push
  pushing to $TESTTMP/remote
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 2 changesets with 2 changes to 2 files (+1 heads)


Remote head is unknown but obsoleted by a local changeset
=========================================================

setup

  $ rm -fr ../remote
  $ cp -R ../backup1 ../remote
  $ cd ..
  $ rm -rf local
  $ hg clone remote local -r 0
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files
  new changesets b4952fcf48cf
  updating to branch default
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ cd local
  $ mkcommit new
  $ hg -R ../remote id --debug -r tip
  c70b08862e0838ea6d7c59c85da2f1ed6c8d67da tip
  $ hg  id --debug -r tip
  71e3228bffe1886550777233d6c97bb5a6b2a650 tip
  $ hg debugobsolete c70b08862e0838ea6d7c59c85da2f1ed6c8d67da 71e3228bffe1886550777233d6c97bb5a6b2a650
  1 new obsolescence markers
  $ hg log -G --hidden
  @  71e3228bffe1 (draft) add new
  |
  o  b4952fcf48cf (public) add base
  
  $ hg log -G --hidden -R ../remote
  o  c70b08862e08 (draft) add old
  |
  @  b4952fcf48cf (public) add base
  

We do not have enought data to take the right decision, we should fail

  $ hg push
  pushing to $TESTTMP/remote
  searching for changes
  remote has heads on branch 'default' that are not known locally: c70b08862e08
  abort: push creates new remote head 71e3228bffe1
  (pull and merge or see 'hg help push' for details about pushing new heads)
  [20]

Pulling the missing data makes it work

  $ hg pull
  pulling from $TESTTMP/remote
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)
  (1 other changesets obsolete on arrival)
  (run 'hg heads' to see heads)
  $ hg push
  pushing to $TESTTMP/remote
  searching for changes
  adding changesets
  adding manifests
  adding file changes
  added 1 changesets with 1 changes to 1 files (+1 heads)

Old head is pruned without parent data and new unrelated head added
===================================================================

setup

  $ cd ..
  $ rm -R remote local
  $ cp -R backup1 remote
  $ hg clone remote local -qr c70b08862e08
  $ cd local
  $ hg up -q '.^'
  $ mkcommit new-unrelated
  created new head
  $ hg debugobsolete `getid old`
  1 new obsolescence markers
  obsoleted 1 changesets
  $ hg log -G --hidden
  @  350a93b716be (draft) add new-unrelated
  |
  | x  c70b08862e08 (draft) add old
  |/
  o  b4952fcf48cf (public) add base
  

  $ hg push
  pushing to $TESTTMP/remote
  searching for changes
  abort: push creates new remote head 350a93b716be
  (merge or see 'hg help push' for details about pushing new heads)
  [20]