view tests/test-racy-mutations.t @ 47802:de2e04fe4897

hgwebdir: avoid systematic full garbage collection

Forcing a systematic full garbage collection upon each request
can serioulsy harm performance. This is reported as
https://bz.mercurial-scm.org/show_bug.cgi?id=6075

With this change we're performing the full collection according
to a new setting, `experimental.web.full-garbage-collection-rate`.
The default value is 1, which doesn't change the behavior and will
allow us to test on real use cases. If the value is 0, no full garbage
collection occurs.

Regardless of the value of the setting, a partial garbage collection
still occurs upon each request (not attempting to collect objects from
the oldest generation). This should be enough to take care of
reference cycles that have been created by the last request
(assessment of this requires changing the setting, not to be 1).

In my experience chasing memory leaks in Mercurial servers,
the full collection never reclaimed any memory, but this is with
Python 3 and biased towards small repositories.

On the other hand, as explained in the Python developer docs [1],
frequent full collections are very harmful in terms of performance if
lots of objects survive the collection, and hence stay in the
oldest generation. Note that `gc.collect()` is indeed trying to
collect the oldest generation [2]. This happens usually in two cases:

- unwanted lingering objects (i.e., an actual memory leak that
  the GC cannot do anything about). Sadly, we have lots of those
  these days.
- desireable long-term objects, typically in caches (not inner caches
  carried by repositories, which should be collected with them). This
  is a subject of interest for the Heptapod project.

In short, the flat rate that this change still permits is
probably a bad idea in most cases, and the default value can
be tweaked later on (or even be set to 0) according to experiments
in the wild.

The test is inspired from test-hgwebdir-paths.py

[1] https://devguide.python.org/garbage_collector/#collecting-the-oldest-generation
[2] https://docs.python.org/3/library/gc.html#gc.collect

Differential Revision: https://phab.mercurial-scm.org/D11204

#testcases skip-detection fail-if-detected

Test situations that "should" only be reproducible:
- on networked filesystems, or
- user using `hg debuglocks` to eliminate the lock file, or
- something (that doesn't respect the lock file) writing to the .hg directory
while we're running

  $ hg init a
  $ cd a

  $ cat > "$TESTTMP/waitlock_editor.sh" <<EOF
  >     [ -n "\${WAITLOCK_ANNOUNCE:-}" ] && touch "\${WAITLOCK_ANNOUNCE}"
  >     f="\${WAITLOCK_FILE}"
  >     start=\`date +%s\`
  >     timeout=5
  >     while [ \\( ! -f \$f \\) -a \\( ! -L \$f \\) ]; do
  >         now=\`date +%s\`
  >         if [ "\`expr \$now - \$start\`" -gt \$timeout ]; then
  >             echo "timeout: \$f was not created in \$timeout seconds (it is now \$(date +%s))"
  >             exit 1
  >         fi
  >         sleep 0.1
  >     done
  >     if [ \$# -gt 1 ]; then
  >         cat "\$@"
  >     fi
  > EOF

Things behave differently if we don't already have a 00changelog.i file when
this all starts, so let's make one.

  $ echo r0 > r0
  $ hg commit -qAm 'r0'

Start an hg commit that will take a while
  $ EDITOR_STARTED="$(pwd)/.editor_started"
  $ MISCHIEF_MANAGED="$(pwd)/.mischief_managed"
  $ JOBS_FINISHED="$(pwd)/.jobs_finished"

#if fail-if-detected
  $ cat >> .hg/hgrc << EOF
  > [debug]
  > revlog.verifyposition.changelog = fail
  > EOF
#endif

  $ echo foo > foo
  $ (WAITLOCK_ANNOUNCE="${EDITOR_STARTED}" \
  >      WAITLOCK_FILE="${MISCHIEF_MANAGED}" \
  >           HGEDITOR="sh $TESTTMP/waitlock_editor.sh" \
  >           hg commit -qAm 'r1 (foo)' --edit foo > .foo_commit_out 2>&1 ; touch "${JOBS_FINISHED}") &

Wait for the "editor" to actually start
  $ WAITLOCK_FILE="${EDITOR_STARTED}" sh "$TESTTMP/waitlock_editor.sh"

Break the locks, and make another commit.
  $ hg debuglocks -LW
  $ echo bar > bar
  $ hg commit -qAm 'r2 (bar)' bar
  $ hg debugrevlogindex -c
     rev linkrev nodeid       p1           p2
       0       0 222799e2f90b 000000000000 000000000000
       1       1 6f124f6007a0 222799e2f90b 000000000000

Awaken the editor from that first commit
  $ touch "${MISCHIEF_MANAGED}"
And wait for it to finish
  $ WAITLOCK_FILE="${JOBS_FINISHED}" sh "$TESTTMP/waitlock_editor.sh"

#if skip-detection
(Ensure there was no output)
  $ cat .foo_commit_out
And observe a corrupted repository -- rev 2's linkrev is 1, which should never
happen for the changelog (the linkrev should always refer to itself).
  $ hg debugrevlogindex -c
     rev linkrev nodeid       p1           p2
       0       0 222799e2f90b 000000000000 000000000000
       1       1 6f124f6007a0 222799e2f90b 000000000000
       2       1 ac80e6205bb2 222799e2f90b 000000000000
#endif

#if fail-if-detected
  $ cat .foo_commit_out
  transaction abort!
  rollback completed
  note: commit message saved in .hg/last-message.txt
  note: use 'hg commit --logfile .hg/last-message.txt --edit' to reuse it
  abort: 00changelog.i: file cursor at position 249, expected 121
And no corruption in the changelog.
  $ hg debugrevlogindex -c
     rev linkrev nodeid       p1           p2
       0       0 222799e2f90b 000000000000 000000000000
       1       1 6f124f6007a0 222799e2f90b 000000000000 (missing-correct-output !)
And, because of transactions, there's none in the manifestlog either.
  $ hg debugrevlogindex -m
     rev linkrev nodeid       p1           p2
       0       0 7b7020262a56 000000000000 000000000000
       1       1 ad3fe36d86d9 7b7020262a56 000000000000
#endif