Mercurial > hg
view mercurial/destutil.py @ 39506:b66ea3fc3a86
sparse-revlog: set max delta chain length to on thousand
The new snapshot system used in the sparse-revlog case gave us some small size
benefit so far. However its most important property is to gracefully handle
harder limit on delta chainlength.
Long delta chain has a very detrimental impact on read (and write) performance
in revlog. Being able to shorter them provide a great boost. However, shorting
delta used to result significantly lower compression ratio. The intermediate
snapshots effectively suppress most of this effect (even all in some case).
# Effect on the test repository
The repository we use for test is not "realistic" but can still show this in
action using an unreasonably low chain limit. Limiting the chain length show a
sizeable increase but stay under control: +6% for limit=15; +15% for limit=10.
Without the snapshot system the increase is significantly bigger: +45% for
limit=15; +80% for limit=10. Even slightly larger than without delta chain
limit, the resulting size is still smaller than before we started doing
snapshots.
Here is a table for comparison. *Since the repository is not branchy, the
initial sparse-revlog version does not bring much benefit compare to the
non-sparse one):
chain length limit | none | limit=15 | limit=10 |
without sparse-revlog | 62 818 987 | 112 664 615 | 131 222 574 |
without snapshot | 74 365 490 | 108 211 410 | 133 857 764 |
with snapshot | 59 230 936 | 63 002 924 | 68 415 329 |
# Effect On Real Life Repositories
The series provides significant benefits on all kind of repositories.
Using `hg debugupgraderepo -o redeltaparent --run`, we recomputed delta chain
for various repositories with different settings:
- delta chain length: unlimited or 1000 limit
- sparse-revlog: enabled or disabled
- this series: applied or not applied
We can observe multiple types of effect:
- On very branchy repositories:
* The delta chain limit as low impact on the repo size.
* Intermediate snapshot greatly reduces manifest size:
- pypy: -80%
- netbeans: -95%
* The delta chain limit is effective, without a size impact:
- netbeans average: 613 -> 282
- private #1 average: 1 068 -> 307
- On more linear repository:
* Intermediate snapshot limit the impact of delta chain limit:
- mozilla:
without the series: +360%
with the series: +25%
* The delta chain limit provides large improvement:
- mozilla's average chain length:
unlimited: 15 338
limited: 469
* Despite the chain length limit, the manifest size is reduced:
- mercurial: -25%
- mozilla: -30%
It is clear that the use of chains of intermediate snapshots provide large
benefits both in storage size and delta chains quality. We should now switch our
effort toward making sure the write performance are acceptable. Then,
`sparse-revlog` will be a suitable format for all new repository.
# Raw Statistic
* no-sparse: general delta repository not using sparse-revlog
* no-snapshot: sparse-revlog repository not using this series
* snapshot: sparse-revlog repository using this series
mercurial
Manifest Size:
limit | none | 1000
------------|-------------|------------
no-sparse | 8 021 373 | 8 199 366
no-snapshot | 8 103 561 | 8 259 719
snapshot | 6 137 116 | 6 126 433
Manifest Chain length data
limit || none || 1000 ||
value || average | max || average | max ||
------------||---------|---------||---------|---------||
no-sparse || 307 | 1456 || 279 | 1000 ||
no-snapshot || 312 | 1456 || 283 | 1000 ||
snapshot || 248 | 1208 || 241 | 1000 ||
Full Store Size
limit | none | 1000
------------|-------------|------------
no-sparse | 51 013 198 | 51 201 574
no-snapshot | 50 930 795 | 51 141 006
snapshot | 48 072 037 | 48 093 572
pypy
Manifest Size:
limit | none | 1000
------------|-------------|------------
no-sparse | 193 987 784 | 193 987 784
no-snapshot | 163 171 745 | 163 312 229
snapshot | 34 605 900 | 34 600 750
Manifest Chain length data
limit || none || 1000 ||
value || average | max || average | max ||
------------||---------|---------||---------|---------||
no-sparse || 101 | 692 || 101 | 692 ||
no-snapshot || 151 | 1307 || 148 | 1000 ||
snapshot || 128 | 1309 || 125 | 1000 ||
Full Store Size
limit | none | 1000
------------|-------------|------------
no-sparse | 495 931 473 | 495 931 473
no-snapshot | 465 441 017 | 465 581 501
snapshot | 355 467 301 | 355 472 451
Mozilla
Manifest Size:
limit | none | 1000
------------|----------------|---------------
no-sparse | 416 757 148 | 1 869 009 668
no-snapshot | 401 592 370 | 1 843 493 795
snapshot | 224 359 521 | 284 615 500
Manifest Chain length data
limit || none || 1000 ||
value || average | max || average | max ||
------------||---------|---------||---------|---------||
no-sparse || 15 333 | 58 980 || 468 | 1 000 ||
no-snapshot || 15 336 | 58 980 || 469 | 1 000 ||
snapshot || 15 338 | 58 983 || 469 | 1 000 ||
Full Store Size
limit | none | 1000
------------|----------------|---------------
no-sparse | 2 712 477 887 | 4 164 995 451
no-snapshot | 2 698 887 835 | 4 141 054 304
snapshot | 2 518 130 385 | 2 578 587 596
Netbeans
Manifest Size:
limit | none | 1000
------------|----------------|---------------
no-sparse | 4 766 794 101 | 4 870 642 687
no-snapshot | 4 334 806 082 | 4 428 681 309
snapshot | 232 659 666 | 240 330 665
Manifest Chain length data
limit || none || 1000 ||
value || average | max || average | max ||
------------||---------|---------||---------|---------||
no-sparse || 597 | 6802 || 254 | 1 000 ||
no-snapshot || 648 | 6 802 || 305 | 1 000 ||
snapshot || 613 | 6 804 || 282 | 1 000 ||
Full Store Size
limit | none | 1000
------------|----------------|---------------
no-sparse | 5 807 347 998 | 5 911 196 584
no-snapshot | 5 375 398 602 | 5 469 273 829
snapshot | 1 282 519 928 | 1 290 190 927
Private repo #1
Manifest Size:
limit | none | 1000
------------|-----------------|---------------
no-sparse | 41 389 010 840 | 41 398 162 091
no-snapshot | 9 737 319 435 | 10 223 773 150
snapshot | 744 215 807 | 747 961 822
Manifest Chain length data
limit || none || 1000 ||
value || average | max || average | max ||
------------||---------|---------||---------|---------||
no-sparse || 245 | 8 885 || 81 | 1 000 ||
no-snapshot || 1 225 | 8 885 || 336 | 1 000 ||
snapshot || 1 068 | 7 909 || 307 | 1 000 ||
Full Store Size
limit | none | 1000
------------|----------------|---------------
no-sparse | 49 646 065 126 | 49 655 216 377
no-snapshot | 17 924 862 856 | 18 411 316 571
snapshot | 9 009 024 710 | 9 012 770 725
Private repo #2
We currently have less data available for that repository.
* Before is a sparse-revlog repository without this series
* After is a sparse-revlog repository with this series + 1000 chain limit
Manifest Size:
Before: 1 531 485 040 bytes
After: 1 091 422 451 bytes
Manifest Chain:
Before: 2 218 avg; 6 575 Max
After: 442 avg; 1 000 Max
Full Store Size
Before: 15 203 955 615
after: 8 207 180 693
author | Boris Feld <boris.feld@octobus.net> |
---|---|
date | Fri, 07 Sep 2018 11:18:45 -0400 |
parents | 92213f6745ed |
children | fbd168455b26 |
line wrap: on
line source
# destutil.py - Mercurial utility function for command destination # # Copyright Matt Mackall <mpm@selenic.com> and other # # This software may be used and distributed according to the terms of the # GNU General Public License version 2 or any later version. from __future__ import absolute_import from .i18n import _ from . import ( bookmarks, error, obsutil, scmutil, stack ) def orphanpossibledestination(repo, rev): """Return all changesets that may be a new parent for orphan `rev`. This function works fine on non-orphan revisions, it's just silly because there's no destination implied by obsolete markers, so it'll return nothing. """ tonode = repo.changelog.node parents = repo.changelog.parentrevs torev = repo.changelog.rev dest = set() tovisit = list(parents(rev)) while tovisit: r = tovisit.pop() succsets = obsutil.successorssets(repo, tonode(r)) if not succsets: # if there are no successors for r, r was probably pruned # and we should walk up to r's parents to try and find # some successors. tovisit.extend(parents(r)) else: # We should probably pick only one destination from split # (case where '1 < len(ss)'), This could be the currently # tipmost, but the correct result is less clear when # results of the split have been moved such that they # reside on multiple branches. for ss in succsets: for n in ss: dr = torev(n) if dr != -1: dest.add(dr) return dest def _destupdateobs(repo, clean): """decide of an update destination from obsolescence markers""" node = None wc = repo[None] p1 = wc.p1() movemark = None if p1.obsolete() and not p1.children(): # allow updating to successors successors = obsutil.successorssets(repo, p1.node()) # behavior of certain cases is as follows, # # divergent changesets: update to highest rev, similar to what # is currently done when there are more than one head # (i.e. 'tip') # # replaced changesets: same as divergent except we know there # is no conflict # # pruned changeset: no update is done; though, we could # consider updating to the first non-obsolete parent, # similar to what is current done for 'hg prune' if successors: # flatten the list here handles both divergent (len > 1) # and the usual case (len = 1) successors = [n for sub in successors for n in sub] # get the max revision for the given successors set, # i.e. the 'tip' of a set node = repo.revs('max(%ln)', successors).first() if bookmarks.isactivewdirparent(repo): movemark = repo['.'].node() return node, movemark, None def _destupdatebook(repo, clean): """decide on an update destination from active bookmark""" # we also move the active bookmark, if any node = None activemark, movemark = bookmarks.calculateupdate(repo.ui, repo) if activemark is not None: node = repo._bookmarks[activemark] return node, movemark, activemark def _destupdatebranch(repo, clean): """decide on an update destination from current branch This ignores closed branch heads. """ wc = repo[None] movemark = node = None currentbranch = wc.branch() if clean: currentbranch = repo['.'].branch() if currentbranch in repo.branchmap(): heads = repo.branchheads(currentbranch) if heads: node = repo.revs('max(.::(%ln))', heads).first() if bookmarks.isactivewdirparent(repo): movemark = repo['.'].node() elif currentbranch == 'default' and not wc.p1(): # "null" parent belongs to "default" branch, but it doesn't exist, so # update to the tipmost non-closed branch head node = repo.revs('max(head() and not closed())').first() else: node = repo['.'].node() return node, movemark, None def _destupdatebranchfallback(repo, clean): """decide on an update destination from closed heads in current branch""" wc = repo[None] currentbranch = wc.branch() movemark = None if currentbranch in repo.branchmap(): # here, all descendant branch heads are closed heads = repo.branchheads(currentbranch, closed=True) assert heads, "any branch has at least one head" node = repo.revs('max(.::(%ln))', heads).first() assert node is not None, ("any revision has at least " "one descendant branch head") if bookmarks.isactivewdirparent(repo): movemark = repo['.'].node() else: # here, no "default" branch, and all branches are closed node = repo.lookup('tip') assert node is not None, "'tip' exists even in empty repository" return node, movemark, None # order in which each step should be evaluated # steps are run until one finds a destination destupdatesteps = ['evolution', 'bookmark', 'branch', 'branchfallback'] # mapping to ease extension overriding steps. destupdatestepmap = {'evolution': _destupdateobs, 'bookmark': _destupdatebook, 'branch': _destupdatebranch, 'branchfallback': _destupdatebranchfallback, } def destupdate(repo, clean=False): """destination for bare update operation return (rev, movemark, activemark) - rev: the revision to update to, - movemark: node to move the active bookmark from (cf bookmark.calculate update), - activemark: a bookmark to activate at the end of the update. """ node = movemark = activemark = None for step in destupdatesteps: node, movemark, activemark = destupdatestepmap[step](repo, clean) if node is not None: break rev = repo[node].rev() return rev, movemark, activemark msgdestmerge = { # too many matching divergent bookmark 'toomanybookmarks': {'merge': (_("multiple matching bookmarks to merge -" " please merge with an explicit rev or bookmark"), _("run 'hg heads' to see all heads")), 'rebase': (_("multiple matching bookmarks to rebase -" " please rebase to an explicit rev or bookmark"), _("run 'hg heads' to see all heads")), }, # no other matching divergent bookmark 'nootherbookmarks': {'merge': (_("no matching bookmark to merge - " "please merge with an explicit rev or bookmark"), _("run 'hg heads' to see all heads")), 'rebase': (_("no matching bookmark to rebase - " "please rebase to an explicit rev or bookmark"), _("run 'hg heads' to see all heads")), }, # branch have too many unbookmarked heads, no obvious destination 'toomanyheads': {'merge': (_("branch '%s' has %d heads - please merge with an explicit rev"), _("run 'hg heads .' to see heads")), 'rebase': (_("branch '%s' has %d heads - please rebase to an explicit rev"), _("run 'hg heads .' to see heads")), }, # branch have no other unbookmarked heads 'bookmarkedheads': {'merge': (_("heads are bookmarked - please merge with an explicit rev"), _("run 'hg heads' to see all heads")), 'rebase': (_("heads are bookmarked - please rebase to an explicit rev"), _("run 'hg heads' to see all heads")), }, # branch have just a single heads, but there is other branches 'nootherbranchheads': {'merge': (_("branch '%s' has one head - please merge with an explicit rev"), _("run 'hg heads' to see all heads")), 'rebase': (_("branch '%s' has one head - please rebase to an explicit rev"), _("run 'hg heads' to see all heads")), }, # repository have a single head 'nootherheads': {'merge': (_('nothing to merge'), None), 'rebase': (_('nothing to rebase'), None), }, # repository have a single head and we are not on it 'nootherheadsbehind': {'merge': (_('nothing to merge'), _("use 'hg update' instead")), 'rebase': (_('nothing to rebase'), _("use 'hg update' instead")), }, # We are not on a head 'notatheads': {'merge': (_('working directory not at a head revision'), _("use 'hg update' or merge with an explicit revision")), 'rebase': (_('working directory not at a head revision'), _("use 'hg update' or rebase to an explicit revision")) }, 'emptysourceset': {'merge': (_('source set is empty'), None), 'rebase': (_('source set is empty'), None), }, 'multiplebranchessourceset': {'merge': (_('source set is rooted in multiple branches'), None), 'rebase': (_('rebaseset is rooted in multiple named branches'), _('specify an explicit destination with --dest')), }, } def _destmergebook(repo, action='merge', sourceset=None, destspace=None): """find merge destination in the active bookmark case""" node = None bmheads = bookmarks.headsforactive(repo) curhead = repo._bookmarks[repo._activebookmark] if len(bmheads) == 2: if curhead == bmheads[0]: node = bmheads[1] else: node = bmheads[0] elif len(bmheads) > 2: msg, hint = msgdestmerge['toomanybookmarks'][action] raise error.ManyMergeDestAbort(msg, hint=hint) elif len(bmheads) <= 1: msg, hint = msgdestmerge['nootherbookmarks'][action] raise error.NoMergeDestAbort(msg, hint=hint) assert node is not None return node def _destmergebranch(repo, action='merge', sourceset=None, onheadcheck=True, destspace=None): """find merge destination based on branch heads""" node = None if sourceset is None: sourceset = [repo[repo.dirstate.p1()].rev()] branch = repo.dirstate.branch() elif not sourceset: msg, hint = msgdestmerge['emptysourceset'][action] raise error.NoMergeDestAbort(msg, hint=hint) else: branch = None for ctx in repo.set('roots(%ld::%ld)', sourceset, sourceset): if branch is not None and ctx.branch() != branch: msg, hint = msgdestmerge['multiplebranchessourceset'][action] raise error.ManyMergeDestAbort(msg, hint=hint) branch = ctx.branch() bheads = repo.branchheads(branch) onhead = repo.revs('%ld and %ln', sourceset, bheads) if onheadcheck and not onhead: # Case A: working copy if not on a head. (merge only) # # This is probably a user mistake We bailout pointing at 'hg update' if len(repo.heads()) <= 1: msg, hint = msgdestmerge['nootherheadsbehind'][action] else: msg, hint = msgdestmerge['notatheads'][action] raise error.Abort(msg, hint=hint) # remove heads descendants of source from the set bheads = list(repo.revs('%ln - (%ld::)', bheads, sourceset)) # filters out bookmarked heads nbhs = list(repo.revs('%ld - bookmark()', bheads)) if destspace is not None: # restrict search space # used in the 'hg pull --rebase' case, see issue 5214. nbhs = list(repo.revs('%ld and %ld', destspace, nbhs)) if len(nbhs) > 1: # Case B: There is more than 1 other anonymous heads # # This means that there will be more than 1 candidate. This is # ambiguous. We abort asking the user to pick as explicit destination # instead. msg, hint = msgdestmerge['toomanyheads'][action] msg %= (branch, len(bheads) + 1) raise error.ManyMergeDestAbort(msg, hint=hint) elif not nbhs: # Case B: There is no other anonymous heads # # This means that there is no natural candidate to merge with. # We abort, with various messages for various cases. if bheads: msg, hint = msgdestmerge['bookmarkedheads'][action] elif len(repo.heads()) > 1: msg, hint = msgdestmerge['nootherbranchheads'][action] msg %= branch elif not onhead: # if 'onheadcheck == False' (rebase case), # this was not caught in Case A. msg, hint = msgdestmerge['nootherheadsbehind'][action] else: msg, hint = msgdestmerge['nootherheads'][action] raise error.NoMergeDestAbort(msg, hint=hint) else: node = nbhs[0] assert node is not None return node def destmerge(repo, action='merge', sourceset=None, onheadcheck=True, destspace=None): """return the default destination for a merge (or raise exception about why it can't pick one) :action: the action being performed, controls emitted error message """ # destspace is here to work around issues with `hg pull --rebase` see # issue5214 for details if repo._activebookmark: node = _destmergebook(repo, action=action, sourceset=sourceset, destspace=destspace) else: node = _destmergebranch(repo, action=action, sourceset=sourceset, onheadcheck=onheadcheck, destspace=destspace) return repo[node].rev() def desthistedit(ui, repo): """Default base revision to edit for `hg histedit`.""" default = ui.config('histedit', 'defaultrev') if default is None: revs = stack.getstack(repo) elif default: revs = scmutil.revrange(repo, [default]) if revs: # Take the first revision of the revset as the root return revs.min() return None def stackbase(ui, repo): revs = stack.getstack(repo) return revs.first() if revs else None def _statusotherbook(ui, repo): bmheads = bookmarks.headsforactive(repo) curhead = repo._bookmarks[repo._activebookmark] if repo.revs('%n and parents()', curhead): # we are on the active bookmark bmheads = [b for b in bmheads if curhead != b] if bmheads: msg = _('%i other divergent bookmarks for "%s"\n') ui.status(msg % (len(bmheads), repo._activebookmark)) def _statusotherbranchheads(ui, repo): currentbranch = repo.dirstate.branch() allheads = repo.branchheads(currentbranch, closed=True) heads = repo.branchheads(currentbranch) if repo.revs('%ln and parents()', allheads): # we are on a head, even though it might be closed # # on closed otherheads # ========= ========== # o 0 all heads for current branch are closed # N only descendant branch heads are closed # x 0 there is only one non-closed branch head # N there are some non-closed branch heads # ========= ========== otherheads = repo.revs('%ln - parents()', heads) if repo['.'].closesbranch(): ui.warn(_('no open descendant heads on branch "%s", ' 'updating to a closed head\n') % (currentbranch)) if otherheads: ui.warn(_("(committing will reopen the head, " "use 'hg heads .' to see %i other heads)\n") % (len(otherheads))) else: ui.warn(_('(committing will reopen branch "%s")\n') % (currentbranch)) elif otherheads: curhead = repo['.'] ui.status(_('updated to "%s: %s"\n') % (curhead, curhead.description().split('\n')[0])) ui.status(_('%i other heads for branch "%s"\n') % (len(otherheads), currentbranch)) def statusotherdests(ui, repo): """Print message about other head""" # XXX we should probably include a hint: # - about what to do # - how to see such heads if repo._activebookmark: _statusotherbook(ui, repo) else: _statusotherbranchheads(ui, repo)