Mercurial > hg
view tests/test-absorb-rename.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.
author | Georges Racinet <georges.racinet@octobus.net> |
---|---|
date | Fri, 23 Apr 2021 18:30:53 +0200 |
parents | 31dfa7dac4c9 |
children |
line wrap: on
line source
$ cat >> $HGRCPATH << EOF > [diff] > git=1 > [extensions] > absorb= > EOF $ sedi() { # workaround check-code > pattern="$1" > shift > for i in "$@"; do > sed "$pattern" "$i" > "$i".tmp > mv "$i".tmp "$i" > done > } rename a to b, then b to a $ hg init repo1 $ cd repo1 $ echo 1 > a $ hg ci -A a -m 1 $ hg mv a b $ echo 2 >> b $ hg ci -m 2 $ hg mv b a $ echo 3 >> a $ hg ci -m 3 $ hg annotate -ncf a 0 eff892de26ec a: 1 1 bf56e1f4f857 b: 2 2 0b888b00216c a: 3 $ sedi 's/$/a/' a $ hg absorb -apq showing changes for a @@ -0,3 +0,3 @@ eff892d -1 bf56e1f -2 0b888b0 -3 eff892d +1a bf56e1f +2a 0b888b0 +3a 3 changesets affected 0b888b0 3 bf56e1f 2 eff892d 1 $ hg status $ hg annotate -ncf a 0 5d1c5620e6f2 a: 1a 1 9a14ffe67ae9 b: 2a 2 9191d121a268 a: 3a when the first changeset is public $ hg phase --public -r 0 $ sedi 's/a/A/' a $ hg absorb -apq showing changes for a @@ -0,3 +0,3 @@ -1a 9a14ffe -2a 9191d12 -3a +1A 9a14ffe +2A 9191d12 +3A 2 changesets affected 9191d12 3 9a14ffe 2 $ hg diff diff --git a/a b/a --- a/a +++ b/a @@ -1,3 +1,3 @@ -1a +1A 2A 3A copy a to b $ cd .. $ hg init repo2 $ cd repo2 $ echo 1 > a $ hg ci -A a -m 1 $ hg cp a b $ echo 2 >> b $ hg ci -m 2 $ hg log -T '{rev}:{node|short} {desc}\n' 1:17b72129ab68 2 0:eff892de26ec 1 $ sedi 's/$/a/' a $ sedi 's/$/b/' b $ hg absorb -apq showing changes for a @@ -0,1 +0,1 @@ eff892d -1 eff892d +1a showing changes for b @@ -0,2 +0,2 @@ -1 17b7212 -2 +1b 17b7212 +2b 2 changesets affected 17b7212 2 eff892d 1 $ hg diff diff --git a/b b/b --- a/b +++ b/b @@ -1,2 +1,2 @@ -1 +1b 2b copy b to a $ cd .. $ hg init repo3 $ cd repo3 $ echo 1 > b $ hg ci -A b -m 1 $ hg cp b a $ echo 2 >> a $ hg ci -m 2 $ hg log -T '{rev}:{node|short} {desc}\n' 1:e62c256d8b24 2 0:55105f940d5c 1 $ sedi 's/$/a/' a $ sedi 's/$/a/' b $ hg absorb -apq showing changes for a @@ -0,2 +0,2 @@ -1 e62c256 -2 +1a e62c256 +2a showing changes for b @@ -0,1 +0,1 @@ 55105f9 -1 55105f9 +1a 2 changesets affected e62c256 2 55105f9 1 $ hg diff diff --git a/a b/a --- a/a +++ b/a @@ -1,2 +1,2 @@ -1 +1a 2a "move" b to both a and c, follow a - sorted alphabetically $ cd .. $ hg init repo4 $ cd repo4 $ echo 1 > b $ hg ci -A b -m 1 $ hg cp b a $ hg cp b c $ hg rm b $ echo 2 >> a $ echo 3 >> c $ hg commit -m cp $ hg log -T '{rev}:{node|short} {desc}\n' 1:366daad8e679 cp 0:55105f940d5c 1 $ sedi 's/$/a/' a $ sedi 's/$/c/' c $ hg absorb -apq showing changes for a @@ -0,2 +0,2 @@ 55105f9 -1 366daad -2 55105f9 +1a 366daad +2a showing changes for c @@ -0,2 +0,2 @@ -1 366daad -3 +1c 366daad +3c 2 changesets affected 366daad cp 55105f9 1 $ hg log -G -p -T '{rev}:{node|short} {desc}\n' @ 1:70606019f91b cp | diff --git a/b b/a | rename from b | rename to a | --- a/b | +++ b/a | @@ -1,1 +1,2 @@ | 1a | +2a | diff --git a/b b/c | copy from b | copy to c | --- a/b | +++ b/c | @@ -1,1 +1,2 @@ | -1a | +1 | +3c | o 0:bfb67c3539c1 1 diff --git a/b b/b new file mode 100644 --- /dev/null +++ b/b @@ -0,0 +1,1 @@ +1a run absorb again would apply the change to c $ hg absorb -apq showing changes for c @@ -0,1 +0,1 @@ 7060601 -1 7060601 +1c 1 changesets affected 7060601 cp $ hg log -G -p -T '{rev}:{node|short} {desc}\n' @ 1:8bd536cce368 cp | diff --git a/b b/a | rename from b | rename to a | --- a/b | +++ b/a | @@ -1,1 +1,2 @@ | 1a | +2a | diff --git a/b b/c | copy from b | copy to c | --- a/b | +++ b/c | @@ -1,1 +1,2 @@ | -1a | +1c | +3c | o 0:bfb67c3539c1 1 diff --git a/b b/b new file mode 100644 --- /dev/null +++ b/b @@ -0,0 +1,1 @@ +1a "move" b to a, c and d, follow d if a gets renamed to e, and c is deleted $ cd .. $ hg init repo5 $ cd repo5 $ echo 1 > b $ hg ci -A b -m 1 $ hg cp b a $ hg cp b c $ hg cp b d $ hg rm b $ echo 2 >> a $ echo 3 >> c $ echo 4 >> d $ hg commit -m cp $ hg mv a e $ hg rm c $ hg commit -m mv $ hg log -T '{rev}:{node|short} {desc}\n' 2:49911557c471 mv 1:7bc3d43ede83 cp 0:55105f940d5c 1 $ sedi 's/$/e/' e $ sedi 's/$/d/' d $ hg absorb -apq showing changes for d @@ -0,2 +0,2 @@ 55105f9 -1 7bc3d43 -4 55105f9 +1d 7bc3d43 +4d showing changes for e @@ -0,2 +0,2 @@ -1 7bc3d43 -2 +1e 7bc3d43 +2e 2 changesets affected 7bc3d43 cp 55105f9 1 $ hg diff diff --git a/e b/e --- a/e +++ b/e @@ -1,2 +1,2 @@ -1 +1e 2e $ hg log -G -p -T '{rev}:{node|short} {desc}\n' @ 2:34be9b0c786e mv | diff --git a/c b/c | deleted file mode 100644 | --- a/c | +++ /dev/null | @@ -1,2 +0,0 @@ | -1 | -3 | diff --git a/a b/e | rename from a | rename to e | o 1:13e56db5948d cp | diff --git a/b b/a | rename from b | rename to a | --- a/b | +++ b/a | @@ -1,1 +1,2 @@ | -1d | +1 | +2e | diff --git a/b b/c | copy from b | copy to c | --- a/b | +++ b/c | @@ -1,1 +1,2 @@ | -1d | +1 | +3 | diff --git a/b b/d | copy from b | copy to d | --- a/b | +++ b/d | @@ -1,1 +1,2 @@ | 1d | +4d | o 0:0037613a5dc6 1 diff --git a/b b/b new file mode 100644 --- /dev/null +++ b/b @@ -0,0 +1,1 @@ +1d