changegroup: remove reordering control (BC)
This logic - including the experimental bundle.reorder option -
was originally added in
a8e3931e3fb5 in 2011 and then later ported
to changegroup.py.
The intent of this option and associated logic is to control
the ordering of revisions in deltagroups in changegroups. At the
time it was implemented, only changegroup version 1 existed
and generaldelta revlogs were just coming into the world. Changegroup
version 1 requires that deltas be made against the last revision
sent over the wire. Used with generaldelta, this created an
impedance mismatch of sorts and resulted in changegroup producers
spending a lot of time recomputing deltas.
Revision reordering was introduced so outgoing revisions would be
sent in "generaldelta order" and producers would be able to
reuse internal deltas from storage.
Later on, we introduced changegroup version 2. It supported denoting
which revision a delta was against. So we no longer needed to
sort outgoing revisions to ensure optimal delta generation from the
producer. So, subsequent changegroup versions disabled reordering.
We also later made the changelog not store deltas by default. And
we also made the changelog send out deltas in storage order. Why we
do this for changelog, I'm not sure. Maybe we want to preserve revision
order across clones? It doesn't really matter for this commit.
Fast forward to 2018. We want to abstract storage backends. And having
changegroup code require knowledge about how deltas are stored
internally interferes with that goal.
This commit removes reordering control from changegroup generation.
After this commit, the reordering behavior is:
* The changelog is always sent out in storage order (no behavior
change).
* Non-changelog generaldelta revlogs are reordered to always be in DAG
topological order (previously, generaldelta revlogs would be emitted
in storage order for version 2 and 3 changegroups).
* Non-changelog non-generaldelta revlogs are sent in storage order (no
behavior change).
* There exists no config option to override behavior.
The big difference here is that generaldelta revlogs now *always* have
their revisions sorted in DAG order before going out over the wire. This
behavior was previously only done for changegroup version 1. Version 2
and version 3 changegroups disabled reordering because the interchange
format supported encoding arbitrary delta parents, so reordering wasn't
strictly necessary.
I can think of a few significant implications for this change.
Because changegroup receivers will now see non-changelog revisions
in DAG order instead of storage order, the internal storage order of
manifests and files may differ substantially between producer and
consumer. I don't think this matters that much, since the storage
order of manifests and files is largely hidden from users. Only
the storage order of changelog matters (because `hg log` shows the
changelog in storage order). I don't think there should be any
controversy here.
The reordering of revisions has implications for changegroup producers.
Previously, generaldelta revlogs would be emitted in storage order.
And in the common case, the internally-stored delta could effectively
be copied from disk into the deltagroup delta. This meant that emitting
delta groups for generaldelta revlogs would be mostly linear read I/O.
This is desirable for performance. With us now reordering generaldelta
revlog revisions in DAG order, the read operations may use more random
I/O instead of sequential I/O. This could result in performance
loss. But with the prevalence of SSDs and fast random I/O, I'm not
too worried. (Note: the optimal emission order for revlogs is actually
delta encoding order. But the changegroup code wasn't doing that before
or after this change. We could potentially implement that in a later
commit.)
Changegroups in DAG order will have implications for receivers.
Previously, receiving storage order might mean seeing a number of
interleaved branches. This would mean long delta chains, sparse
I/O, and possibly more fulltext revisions instead of deltas, blowing
up storage storage. (This is the same set of problems that sparse
revlogs aims to address.) With the producer now sending revisions in DAG
order, the receiver also stores revisions in DAG order. That means
revisions for the same DAG branch are all grouped together. And this
should yield better storage outcomes. In other words, sending the
reordered changegroup allows the receiver to have better storage
order and for the producer to not propagate its (possibly sub-optimal)
internal storage order.
On the mozilla-unified repository, this change influences bundle
generation:
$ hg bundle -t none-v2 -a
before: time: real 355.680 secs (user 256.790+0.000 sys 16.820+0.000)
after: time: real 382.950 secs (user 281.700+0.000 sys 17.690+0.000)
before: 7,150,228,967 bytes (uncompressed)
after: 7,041,556,273 bytes (uncompressed)
before: 1,669,063,234 bytes (zstd l=3)
after: 1,628,598,830 bytes (zstd l=3)
$ hg unbundle
before: time: real 511.910 secs (user 466.750+0.000 sys 32.680+0.000)
after: time: real 487.790 secs (user 443.940+0.000 sys 30.840+0.000)
00manifest.d size:
source: 274,924,292 bytes
before: 304,741,626 bytes
after: 245,252,087 bytes
.hg/store total file size:
source: 2,649,133,490
before: 2,680,888,130
after: 2,627,875,673
We see the bundle size drop. That's probably because if a revlog
internally isn't storing a delta, it will choose to delta against
the last emitted revision. And on repos with interleaved branches
(like mozilla-unified), the previous revision could be an unrelated
branch and therefore be a large delta. But with this patch, the
previous revision is likely p1 or p2 and a delta should be small.
We also see the manifest size drop by ~50 MB. It's worth noting that
the manifest actually *increased* in size by ~25 MB in the old
strategy and decreased ~25 MB from its source in the new strategy.
Again, my explanation for this is that the DAG ordering in the
changegroup is resulting in better grouping of revisions in the
receiver, which results in more compact delta chains and higher
storage efficiency.
Unbundle time also dropped. I suspect this is due to the revlog having
to work less to compute deltas since the incoming deltas are more
optimal. i.e. the receiver spends less time resolving fulltext
revisions as incoming deltas bounce around between DAG branches and
delta chains.
We also see bundle generation time increase. This is not desirable.
However, the regression is only significant on the original repository:
if we generate a bundle from the repository created from the new,
always reordered bundles, we're close to baseline (if not at it with
expected noise):
$ hg bundle -t none-v2 -a
before (original): time: real 355.680 secs (user 256.790+0.000 sys 16.820+0.000)
after (original): time: real 382.950 secs (user 281.700+0.000 sys 17.690+0.000)
after (new repo): time: real 362.280 secs (user 260.300+0.000 sys 17.700+0.000)
This regression is a bit worrying because it will impact serving
canonical repositories (that don't have optimal internal storage
unless they are reordered - possibly as part of running
`hg debugupgraderepo`). However, this regression will only be
noticed by very large changegroups. And I'm guessing/hoping that
any repository that large is using clonebundles to mitigate server
load.
Again, sending DAG order isn't the optimal send order for servers:
sending in storage-delta order is. But in order to enable
storage-optimal send order, we'll need a storage API that handles
sorting. Future commits will introduce such an API.
Differential Revision: https://phab.mercurial-scm.org/D4721
Test that qpush cleans things up if it doesn't complete
$ echo "[extensions]" >> $HGRCPATH
$ echo "mq=" >> $HGRCPATH
$ hg init repo
$ cd repo
$ echo foo > foo
$ hg ci -Am 'add foo'
adding foo
$ touch untracked-file
$ echo 'syntax: glob' > .hgignore
$ echo '.hgignore' >> .hgignore
$ hg qinit
test qpush on empty series
$ hg qpush
no patches in series
$ hg qnew patch1
$ echo >> foo
$ hg qrefresh -m 'patch 1'
$ hg qnew patch2
$ echo bar > bar
$ hg add bar
$ hg qrefresh -m 'patch 2'
$ hg qnew --config 'mq.plain=true' -U bad-patch
$ echo >> foo
$ hg qrefresh
$ hg qpop -a
popping bad-patch
popping patch2
popping patch1
patch queue now empty
$ "$PYTHON" -c 'import sys; getattr(sys.stdout, "buffer", sys.stdout).write(b"\xe9\n")' > message
$ cat .hg/patches/bad-patch >> message
$ mv message .hg/patches/bad-patch
$ cat > $TESTTMP/wrapplayback.py <<EOF
> import os
> from mercurial import extensions, transaction
> def wrapplayback(orig,
> journal, report, opener, vfsmap, entries, backupentries,
> unlink=True, checkambigfiles=None):
> orig(journal, report, opener, vfsmap, entries, backupentries, unlink,
> checkambigfiles)
> # Touching files truncated at "transaction.abort" causes
> # forcible re-loading invalidated filecache properties
> # (including repo.changelog)
> for f, o, _ignore in entries:
> if o or not unlink:
> os.utime(opener.join(f), (0.0, 0.0))
> def extsetup(ui):
> extensions.wrapfunction(transaction, '_playback', wrapplayback)
> EOF
$ hg qpush -a --config extensions.wrapplayback=$TESTTMP/wrapplayback.py && echo 'qpush succeeded?!'
applying patch1
applying patch2
applying bad-patch
transaction abort!
rollback completed
cleaning up working directory...
reverting foo
done
abort: decoding near '\xe9': 'ascii' codec can't decode byte 0xe9 in position 0: ordinal not in range(128)! (esc)
[255]
$ hg parents
changeset: 0:bbd179dfa0a7
tag: tip
user: test
date: Thu Jan 01 00:00:00 1970 +0000
summary: add foo
test corrupt status file
$ hg qpush
applying patch1
now at: patch1
$ cp .hg/patches/status .hg/patches/status.orig
$ hg qpop
popping patch1
patch queue now empty
$ cp .hg/patches/status.orig .hg/patches/status
$ hg qpush
abort: working directory revision is not qtip
[255]
$ rm .hg/patches/status .hg/patches/status.orig
bar should be gone; other unknown/ignored files should still be around
$ hg status -A
? untracked-file
I .hgignore
C foo
preparing qpush of a missing patch
$ hg qpop -a
no patches applied
$ hg qpush
applying patch1
now at: patch1
$ rm .hg/patches/patch2
now we expect the push to fail, but it should NOT complain about patch1
$ hg qpush
applying patch2
unable to read patch2
now at: patch1
[1]
preparing qpush of missing patch with no patch applied
$ hg qpop -a
popping patch1
patch queue now empty
$ rm .hg/patches/patch1
qpush should fail the same way as below
$ hg qpush
applying patch1
unable to read patch1
[1]
Test qpush to a patch below the currently applied patch.
$ hg qq -c guardedseriesorder
$ hg qnew a
$ hg qguard +block
$ hg qnew b
$ hg qnew c
$ hg qpop -a
popping c
popping b
popping a
patch queue now empty
try to push and pop while a is guarded
$ hg qpush a
cannot push 'a' - guarded by '+block'
[1]
$ hg qpush -a
applying b
patch b is empty
applying c
patch c is empty
now at: c
now try it when a is unguarded, and we're at the top of the queue
$ hg qapplied -v
0 G a
1 A b
2 A c
$ hg qsel block
$ hg qpush b
abort: cannot push to a previous patch: b
[255]
$ hg qpush a
abort: cannot push to a previous patch: a
[255]
and now we try it one more time with a unguarded, while we're not at the top of the queue
$ hg qpop b
popping c
now at: b
$ hg qpush a
abort: cannot push to a previous patch: a
[255]
test qpop --force and backup files
$ hg qpop -a
popping b
patch queue now empty
$ hg qq --create force
$ echo a > a
$ echo b > b
$ echo c > c
$ hg ci -Am add a b c
$ echo a >> a
$ hg rm b
$ hg rm c
$ hg qnew p1
$ echo a >> a
$ echo bb > b
$ hg add b
$ echo cc > c
$ hg add c
$ hg qpop --force --verbose
saving current version of a as a.orig
saving current version of b as b.orig
saving current version of c as c.orig
popping p1
patch queue now empty
$ hg st
? a.orig
? b.orig
? c.orig
? untracked-file
$ cat a.orig
a
a
a
$ cat b.orig
bb
$ cat c.orig
cc
test qpop --force --no-backup
$ hg qpush
applying p1
now at: p1
$ rm a.orig
$ echo a >> a
$ hg qpop --force --no-backup --verbose
popping p1
patch queue now empty
$ test -f a.orig && echo 'error: backup with --no-backup'
[1]
test qpop --keep-changes
$ hg qpush
applying p1
now at: p1
$ hg qpop --keep-changes --force
abort: cannot use both --force and --keep-changes
[255]
$ echo a >> a
$ hg qpop --keep-changes
abort: local changes found, qrefresh first
[255]
$ hg revert -qa a
$ rm a
$ hg qpop --keep-changes
abort: local changes found, qrefresh first
[255]
$ hg rm -A a
$ hg qpop --keep-changes
abort: local changes found, qrefresh first
[255]
$ hg revert -qa a
$ echo b > b
$ hg add b
$ hg qpop --keep-changes
abort: local changes found, qrefresh first
[255]
$ hg forget b
$ echo d > d
$ hg add d
$ hg qpop --keep-changes
popping p1
patch queue now empty
$ hg forget d
$ rm d
test qpush --force and backup files
$ echo a >> a
$ hg qnew p2
$ echo b >> b
$ echo d > d
$ echo e > e
$ hg add d e
$ hg rm c
$ hg qnew p3
$ hg qpop -a
popping p3
popping p2
patch queue now empty
$ echo a >> a
$ echo b1 >> b
$ echo d1 > d
$ hg add d
$ echo e1 > e
$ hg qpush -a --force --verbose
applying p2
saving current version of a as a.orig
patching file a
committing files:
a
committing manifest
committing changelog
applying p3
saving current version of b as b.orig
saving current version of d as d.orig
patching file b
patching file c
patching file d
file d already exists
1 out of 1 hunks FAILED -- saving rejects to file d.rej
patching file e
file e already exists
1 out of 1 hunks FAILED -- saving rejects to file e.rej
patch failed to apply
committing files:
b
committing manifest
committing changelog
patch failed, rejects left in working directory
errors during apply, please fix and qrefresh p3
[2]
$ cat a.orig
a
a
$ cat b.orig
b
b1
$ cat d.orig
d1
test qpush --force --no-backup
$ hg revert -qa
$ hg qpop -a
popping p3
popping p2
patch queue now empty
$ echo a >> a
$ rm a.orig
$ hg qpush --force --no-backup --verbose
applying p2
patching file a
committing files:
a
committing manifest
committing changelog
now at: p2
$ test -f a.orig && echo 'error: backup with --no-backup'
[1]
test qgoto --force --no-backup
$ hg qpop
popping p2
patch queue now empty
$ echo a >> a
$ hg qgoto --force --no-backup p2 --verbose
applying p2
patching file a
committing files:
a
committing manifest
committing changelog
now at: p2
$ test -f a.orig && echo 'error: backup with --no-backup'
[1]
test qpush --keep-changes
$ hg qpush --keep-changes --force
abort: cannot use both --force and --keep-changes
[255]
$ hg qpush --keep-changes --exact
abort: cannot use --exact and --keep-changes together
[255]
$ echo b >> b
$ hg qpush --keep-changes
applying p3
abort: conflicting local changes found
(did you forget to qrefresh?)
[255]
$ rm b
$ hg qpush --keep-changes
applying p3
abort: conflicting local changes found
(did you forget to qrefresh?)
[255]
$ hg rm -A b
$ hg qpush --keep-changes
applying p3
abort: conflicting local changes found
(did you forget to qrefresh?)
[255]
$ hg revert -aq b
$ echo d > d
$ hg add d
$ hg qpush --keep-changes
applying p3
abort: conflicting local changes found
(did you forget to qrefresh?)
[255]
$ hg forget d
$ rm d
$ hg qpop
popping p2
patch queue now empty
$ echo b >> b
$ hg qpush -a --keep-changes
applying p2
applying p3
abort: conflicting local changes found
(did you forget to qrefresh?)
[255]
$ hg qtop
p2
$ hg parents --template "{rev} {desc}\n"
2 imported patch p2
$ hg st b
M b
$ cat b
b
b
test qgoto --keep-changes
$ hg revert -aq b
$ rm e
$ hg qgoto --keep-changes --force p3
abort: cannot use both --force and --keep-changes
[255]
$ echo a >> a
$ hg qgoto --keep-changes p3
applying p3
now at: p3
$ hg st a
M a
$ hg qgoto --keep-changes p2
popping p3
now at: p2
$ hg st a
M a
test mq.keepchanges setting
$ hg --config mq.keepchanges=1 qpush
applying p3
now at: p3
$ hg st a
M a
$ hg --config mq.keepchanges=1 qpop
popping p3
now at: p2
$ hg st a
M a
$ hg --config mq.keepchanges=1 qgoto p3
applying p3
now at: p3
$ hg st a
M a
$ echo b >> b
$ hg --config mq.keepchanges=1 qpop --force --config 'ui.origbackuppath=.hg/origbackups'
popping p3
now at: p2
$ hg st b
$ hg --config mq.keepchanges=1 qpush --exact
abort: local changes found, qrefresh first
[255]
$ hg revert -qa a
$ hg qpop
popping p2
patch queue now empty
$ echo a >> a
$ hg --config mq.keepchanges=1 qpush --force
applying p2
now at: p2
$ hg st a
test previous qpop (with --force and --config) saved .orig files to where user
wants them
$ ls .hg/origbackups
b
$ rm -rf .hg/origbackups
$ cd ..