git/t/t6500-gc.sh
Jeff King c45af94dbc gc: run pre-detach operations under lock
We normally try to avoid having two auto-gc operations run
at the same time, because it wastes resources. This was done
long ago in 64a99eb47 (gc: reject if another gc is running,
unless --force is given, 2013-08-08).

When we do a detached auto-gc, we run the ref-related
commands _before_ detaching, to avoid confusing lock
contention. This was done by 62aad1849 (gc --auto: do not
lock refs in the background, 2014-05-25).

These two features do not interact well. The pre-detach
operations are run before we check the gc.pid lock, meaning
that on a busy repository we may run many of them
concurrently. Ideally we'd take the lock before spawning any
operations, and hold it for the duration of the program.

This is tricky, though, with the way the pid-file interacts
with the daemonize() process.  Other processes will check
that the pid recorded in the pid-file still exists. But
detaching causes us to fork and continue running under a
new pid. So if we take the lock before detaching, the
pid-file will have a bogus pid in it. We'd have to go back
and update it with the new pid after detaching. We'd also
have to play some tricks with the tempfile subsystem to
tweak the "owner" field, so that the parent process does not
clean it up on exit, but the child process does.

Instead, we can do something a bit simpler: take the lock
only for the duration of the pre-detach work, then detach,
then take it again for the post-detach work. Technically,
this means that the post-detach lock could lose to another
process doing pre-detach work. But in the long run this
works out.

That second process would then follow-up by doing
post-detach work. Unless it was in turn blocked by a third
process doing pre-detach work, and so on. This could in
theory go on indefinitely, as the pre-detach work does not
repack, and so need_to_gc() will continue to trigger.  But
in each round we are racing between the pre- and post-detach
locks. Eventually, one of the post-detach locks will win the
race and complete the full gc. So in the worst case, we may
racily repeat the pre-detach work, but we would never do so
simultaneously (it would happen via a sequence of serialized
race-wins).

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2017-07-12 09:41:04 -07:00

124 lines
3.6 KiB
Bash
Executable file

#!/bin/sh
test_description='basic git gc tests
'
. ./test-lib.sh
test_expect_success 'gc empty repository' '
git gc
'
test_expect_success 'gc does not leave behind pid file' '
git gc &&
test_path_is_missing .git/gc.pid
'
test_expect_success 'gc --gobbledegook' '
test_expect_code 129 git gc --nonsense 2>err &&
test_i18ngrep "[Uu]sage: git gc" err
'
test_expect_success 'gc -h with invalid configuration' '
mkdir broken &&
(
cd broken &&
git init &&
echo "[gc] pruneexpire = CORRUPT" >>.git/config &&
test_expect_code 129 git gc -h >usage 2>&1
) &&
test_i18ngrep "[Uu]sage" broken/usage
'
test_expect_success 'gc is not aborted due to a stale symref' '
git init remote &&
(
cd remote &&
test_commit initial &&
git clone . ../client &&
git branch -m develop &&
cd ../client &&
git fetch --prune &&
git gc
)
'
test_expect_success 'auto gc with too many loose objects does not attempt to create bitmaps' '
test_config gc.auto 3 &&
test_config gc.autodetach false &&
test_config pack.writebitmaps true &&
# We need to create two object whose sha1s start with 17
# since this is what git gc counts. As it happens, these
# two blobs will do so.
test_commit 263 &&
test_commit 410 &&
# Our first gc will create a pack; our second will create a second pack
git gc --auto &&
ls .git/objects/pack | sort >existing_packs &&
test_commit 523 &&
test_commit 790 &&
git gc --auto 2>err &&
test_i18ngrep ! "^warning:" err &&
ls .git/objects/pack/ | sort >post_packs &&
comm -1 -3 existing_packs post_packs >new &&
comm -2 -3 existing_packs post_packs >del &&
test_line_count = 0 del && # No packs are deleted
test_line_count = 2 new # There is one new pack and its .idx
'
run_and_wait_for_auto_gc () {
# We read stdout from gc for the side effect of waiting until the
# background gc process exits, closing its fd 9. Furthermore, the
# variable assignment from a command substitution preserves the
# exit status of the main gc process.
# Note: this fd trickery doesn't work on Windows, but there is no
# need to, because on Win the auto gc always runs in the foreground.
doesnt_matter=$(git gc --auto 9>&1)
}
test_expect_success 'background auto gc does not run if gc.log is present and recent but does if it is old' '
test_commit foo &&
test_commit bar &&
git repack &&
test_config gc.autopacklimit 1 &&
test_config gc.autodetach true &&
echo fleem >.git/gc.log &&
test_must_fail git gc --auto 2>err &&
test_i18ngrep "^error:" err &&
test_config gc.logexpiry 5.days &&
test-chmtime =-345600 .git/gc.log &&
test_must_fail git gc --auto &&
test_config gc.logexpiry 2.days &&
run_and_wait_for_auto_gc &&
ls .git/objects/pack/pack-*.pack >packs &&
test_line_count = 1 packs
'
test_expect_success 'background auto gc respects lock for all operations' '
# make sure we run a background auto-gc
test_commit make-pack &&
git repack &&
test_config gc.autopacklimit 1 &&
test_config gc.autodetach true &&
# create a ref whose loose presence we can use to detect a pack-refs run
git update-ref refs/heads/should-be-loose HEAD &&
test_path_is_file .git/refs/heads/should-be-loose &&
# now fake a concurrent gc that holds the lock; we can use our
# shell pid so that it looks valid.
hostname=$(hostname || echo unknown) &&
printf "$$ %s" "$hostname" >.git/gc.pid &&
# our gc should exit zero without doing anything
run_and_wait_for_auto_gc &&
test_path_is_file .git/refs/heads/should-be-loose
'
# DO NOT leave a detached auto gc process running near the end of the
# test script: it can run long enough in the background to racily
# interfere with the cleanup in 'test_done'.
test_done