Mercurial > hg
view hgext/remotefilelog/repack.py @ 49541:976648e20856
tests: remove non-python3 line matching and tests block
We don't support Python2 anymore
author | Pierre-Yves David <pierre-yves.david@octobus.net> |
---|---|
date | Thu, 22 Sep 2022 16:27:17 +0200 |
parents | d44e3c45f0e4 |
children | d718eddf01d9 |
line wrap: on
line source
import os import time from mercurial.i18n import _ from mercurial.node import short from mercurial import ( encoding, error, lock as lockmod, mdiff, policy, scmutil, util, vfs, ) from mercurial.utils import procutil from . import ( constants, contentstore, datapack, historypack, metadatastore, shallowutil, ) osutil = policy.importmod('osutil') class RepackAlreadyRunning(error.Abort): pass def backgroundrepack(repo, incremental=True, packsonly=False): cmd = [procutil.hgexecutable(), b'-R', repo.origroot, b'repack'] msg = _(b"(running background repack)\n") if incremental: cmd.append(b'--incremental') msg = _(b"(running background incremental repack)\n") if packsonly: cmd.append(b'--packsonly') repo.ui.warn(msg) # We know this command will find a binary, so don't block on it starting. kwargs = {} if repo.ui.configbool(b'devel', b'remotefilelog.bg-wait'): kwargs['record_wait'] = repo.ui.atexit procutil.runbgcommand(cmd, encoding.environ, ensurestart=False, **kwargs) def fullrepack(repo, options=None): """If ``packsonly`` is True, stores creating only loose objects are skipped.""" if util.safehasattr(repo, 'shareddatastores'): datasource = contentstore.unioncontentstore(*repo.shareddatastores) historysource = metadatastore.unionmetadatastore( *repo.sharedhistorystores, allowincomplete=True ) packpath = shallowutil.getcachepackpath( repo, constants.FILEPACK_CATEGORY ) _runrepack( repo, datasource, historysource, packpath, constants.FILEPACK_CATEGORY, options=options, ) if util.safehasattr(repo.manifestlog, 'datastore'): localdata, shareddata = _getmanifeststores(repo) lpackpath, ldstores, lhstores = localdata spackpath, sdstores, shstores = shareddata # Repack the shared manifest store datasource = contentstore.unioncontentstore(*sdstores) historysource = metadatastore.unionmetadatastore( *shstores, allowincomplete=True ) _runrepack( repo, datasource, historysource, spackpath, constants.TREEPACK_CATEGORY, options=options, ) # Repack the local manifest store datasource = contentstore.unioncontentstore( *ldstores, allowincomplete=True ) historysource = metadatastore.unionmetadatastore( *lhstores, allowincomplete=True ) _runrepack( repo, datasource, historysource, lpackpath, constants.TREEPACK_CATEGORY, options=options, ) def incrementalrepack(repo, options=None): """This repacks the repo by looking at the distribution of pack files in the repo and performing the most minimal repack to keep the repo in good shape. """ if util.safehasattr(repo, 'shareddatastores'): packpath = shallowutil.getcachepackpath( repo, constants.FILEPACK_CATEGORY ) _incrementalrepack( repo, repo.shareddatastores, repo.sharedhistorystores, packpath, constants.FILEPACK_CATEGORY, options=options, ) if util.safehasattr(repo.manifestlog, 'datastore'): localdata, shareddata = _getmanifeststores(repo) lpackpath, ldstores, lhstores = localdata spackpath, sdstores, shstores = shareddata # Repack the shared manifest store _incrementalrepack( repo, sdstores, shstores, spackpath, constants.TREEPACK_CATEGORY, options=options, ) # Repack the local manifest store _incrementalrepack( repo, ldstores, lhstores, lpackpath, constants.TREEPACK_CATEGORY, allowincompletedata=True, options=options, ) def _getmanifeststores(repo): shareddatastores = repo.manifestlog.shareddatastores localdatastores = repo.manifestlog.localdatastores sharedhistorystores = repo.manifestlog.sharedhistorystores localhistorystores = repo.manifestlog.localhistorystores sharedpackpath = shallowutil.getcachepackpath( repo, constants.TREEPACK_CATEGORY ) localpackpath = shallowutil.getlocalpackpath( repo.svfs.vfs.base, constants.TREEPACK_CATEGORY ) return ( (localpackpath, localdatastores, localhistorystores), (sharedpackpath, shareddatastores, sharedhistorystores), ) def _topacks(packpath, files, constructor): paths = list(os.path.join(packpath, p) for p in files) packs = list(constructor(p) for p in paths) return packs def _deletebigpacks(repo, folder, files): """Deletes packfiles that are bigger than ``packs.maxpacksize``. Returns ``files` with the removed files omitted.""" maxsize = repo.ui.configbytes(b"packs", b"maxpacksize") if maxsize <= 0: return files # This only considers datapacks today, but we could broaden it to include # historypacks. VALIDEXTS = [b".datapack", b".dataidx"] # Either an oversize index or datapack will trigger cleanup of the whole # pack: oversized = { os.path.splitext(path)[0] for path, ftype, stat in files if (stat.st_size > maxsize and (os.path.splitext(path)[1] in VALIDEXTS)) } for rootfname in oversized: rootpath = os.path.join(folder, rootfname) for ext in VALIDEXTS: path = rootpath + ext repo.ui.debug( b'removing oversize packfile %s (%s)\n' % (path, util.bytecount(os.stat(path).st_size)) ) os.unlink(path) return [row for row in files if os.path.basename(row[0]) not in oversized] def _incrementalrepack( repo, datastore, historystore, packpath, category, allowincompletedata=False, options=None, ): shallowutil.mkstickygroupdir(repo.ui, packpath) files = osutil.listdir(packpath, stat=True) files = _deletebigpacks(repo, packpath, files) datapacks = _topacks( packpath, _computeincrementaldatapack(repo.ui, files), datapack.datapack ) datapacks.extend( s for s in datastore if not isinstance(s, datapack.datapackstore) ) historypacks = _topacks( packpath, _computeincrementalhistorypack(repo.ui, files), historypack.historypack, ) historypacks.extend( s for s in historystore if not isinstance(s, historypack.historypackstore) ) # ``allhistory{files,packs}`` contains all known history packs, even ones we # don't plan to repack. They are used during the datapack repack to ensure # good ordering of nodes. allhistoryfiles = _allpackfileswithsuffix( files, historypack.PACKSUFFIX, historypack.INDEXSUFFIX ) allhistorypacks = _topacks( packpath, (f for f, mode, stat in allhistoryfiles), historypack.historypack, ) allhistorypacks.extend( s for s in historystore if not isinstance(s, historypack.historypackstore) ) _runrepack( repo, contentstore.unioncontentstore( *datapacks, allowincomplete=allowincompletedata ), metadatastore.unionmetadatastore(*historypacks, allowincomplete=True), packpath, category, fullhistory=metadatastore.unionmetadatastore( *allhistorypacks, allowincomplete=True ), options=options, ) def _computeincrementaldatapack(ui, files): opts = { b'gencountlimit': ui.configint(b'remotefilelog', b'data.gencountlimit'), b'generations': ui.configlist(b'remotefilelog', b'data.generations'), b'maxrepackpacks': ui.configint( b'remotefilelog', b'data.maxrepackpacks' ), b'repackmaxpacksize': ui.configbytes( b'remotefilelog', b'data.repackmaxpacksize' ), b'repacksizelimit': ui.configbytes( b'remotefilelog', b'data.repacksizelimit' ), } packfiles = _allpackfileswithsuffix( files, datapack.PACKSUFFIX, datapack.INDEXSUFFIX ) return _computeincrementalpack(packfiles, opts) def _computeincrementalhistorypack(ui, files): opts = { b'gencountlimit': ui.configint( b'remotefilelog', b'history.gencountlimit' ), b'generations': ui.configlist( b'remotefilelog', b'history.generations', [b'100MB'] ), b'maxrepackpacks': ui.configint( b'remotefilelog', b'history.maxrepackpacks' ), b'repackmaxpacksize': ui.configbytes( b'remotefilelog', b'history.repackmaxpacksize', b'400MB' ), b'repacksizelimit': ui.configbytes( b'remotefilelog', b'history.repacksizelimit' ), } packfiles = _allpackfileswithsuffix( files, historypack.PACKSUFFIX, historypack.INDEXSUFFIX ) return _computeincrementalpack(packfiles, opts) def _allpackfileswithsuffix(files, packsuffix, indexsuffix): result = [] fileset = {fn for fn, mode, stat in files} for filename, mode, stat in files: if not filename.endswith(packsuffix): continue prefix = filename[: -len(packsuffix)] # Don't process a pack if it doesn't have an index. if (prefix + indexsuffix) not in fileset: continue result.append((prefix, mode, stat)) return result def _computeincrementalpack(files, opts): """Given a set of pack files along with the configuration options, this function computes the list of files that should be packed as part of an incremental repack. It tries to strike a balance between keeping incremental repacks cheap (i.e. packing small things when possible, and rolling the packs up to the big ones over time). """ limits = list( sorted((util.sizetoint(s) for s in opts[b'generations']), reverse=True) ) limits.append(0) # Group the packs by generation (i.e. by size) generations = [] for i in range(len(limits)): generations.append([]) sizes = {} for prefix, mode, stat in files: size = stat.st_size if size > opts[b'repackmaxpacksize']: continue sizes[prefix] = size for i, limit in enumerate(limits): if size > limit: generations[i].append(prefix) break # Steps for picking what packs to repack: # 1. Pick the largest generation with > gencountlimit pack files. # 2. Take the smallest three packs. # 3. While total-size-of-packs < repacksizelimit: add another pack # Find the largest generation with more than gencountlimit packs genpacks = [] for i, limit in enumerate(limits): if len(generations[i]) > opts[b'gencountlimit']: # Sort to be smallest last, for easy popping later genpacks.extend( sorted(generations[i], reverse=True, key=lambda x: sizes[x]) ) break # Take as many packs from the generation as we can chosenpacks = genpacks[-3:] genpacks = genpacks[:-3] repacksize = sum(sizes[n] for n in chosenpacks) while ( repacksize < opts[b'repacksizelimit'] and genpacks and len(chosenpacks) < opts[b'maxrepackpacks'] ): chosenpacks.append(genpacks.pop()) repacksize += sizes[chosenpacks[-1]] return chosenpacks def _runrepack( repo, data, history, packpath, category, fullhistory=None, options=None ): shallowutil.mkstickygroupdir(repo.ui, packpath) def isold(repo, filename, node): """Check if the file node is older than a limit. Unless a limit is specified in the config the default limit is taken. """ filectx = repo.filectx(filename, fileid=node) filetime = repo[filectx.linkrev()].date() ttl = repo.ui.configint(b'remotefilelog', b'nodettl') limit = time.time() - ttl return filetime[0] < limit garbagecollect = repo.ui.configbool(b'remotefilelog', b'gcrepack') if not fullhistory: fullhistory = history packer = repacker( repo, data, history, fullhistory, category, gc=garbagecollect, isold=isold, options=options, ) with datapack.mutabledatapack(repo.ui, packpath) as dpack: with historypack.mutablehistorypack(repo.ui, packpath) as hpack: try: packer.run(dpack, hpack) except error.LockHeld: raise RepackAlreadyRunning( _( b"skipping repack - another repack " b"is already running" ) ) def keepset(repo, keyfn, lastkeepkeys=None): """Computes a keepset which is not garbage collected. 'keyfn' is a function that maps filename, node to a unique key. 'lastkeepkeys' is an optional argument and if provided the keepset function updates lastkeepkeys with more keys and returns the result. """ if not lastkeepkeys: keepkeys = set() else: keepkeys = lastkeepkeys # We want to keep: # 1. Working copy parent # 2. Draft commits # 3. Parents of draft commits # 4. Pullprefetch and bgprefetchrevs revsets if specified revs = [b'.', b'draft()', b'parents(draft())'] prefetchrevs = repo.ui.config(b'remotefilelog', b'pullprefetch', None) if prefetchrevs: revs.append(b'(%s)' % prefetchrevs) prefetchrevs = repo.ui.config(b'remotefilelog', b'bgprefetchrevs', None) if prefetchrevs: revs.append(b'(%s)' % prefetchrevs) revs = b'+'.join(revs) revs = [b'sort((%s), "topo")' % revs] keep = scmutil.revrange(repo, revs) processed = set() lastmanifest = None # process the commits in toposorted order starting from the oldest for r in reversed(keep._list): if repo[r].p1().rev() in processed: # if the direct parent has already been processed # then we only need to process the delta m = repo[r].manifestctx().readdelta() else: # otherwise take the manifest and diff it # with the previous manifest if one exists if lastmanifest: m = repo[r].manifest().diff(lastmanifest) else: m = repo[r].manifest() lastmanifest = repo[r].manifest() processed.add(r) # populate keepkeys with keys from the current manifest if type(m) is dict: # m is a result of diff of two manifests and is a dictionary that # maps filename to ((newnode, newflag), (oldnode, oldflag)) tuple for filename, diff in m.items(): if diff[0][0] is not None: keepkeys.add(keyfn(filename, diff[0][0])) else: # m is a manifest object for filename, filenode in m.items(): keepkeys.add(keyfn(filename, filenode)) return keepkeys class repacker: """Class for orchestrating the repack of data and history information into a new format. """ def __init__( self, repo, data, history, fullhistory, category, gc=False, isold=None, options=None, ): self.repo = repo self.data = data self.history = history self.fullhistory = fullhistory self.unit = constants.getunits(category) self.garbagecollect = gc self.options = options if self.garbagecollect: if not isold: raise ValueError(b"Function 'isold' is not properly specified") # use (filename, node) tuple as a keepset key self.keepkeys = keepset(repo, lambda f, n: (f, n)) self.isold = isold def run(self, targetdata, targethistory): ledger = repackledger() with lockmod.lock( repacklockvfs(self.repo), b"repacklock", desc=None, timeout=0 ): self.repo.hook(b'prerepack') # Populate ledger from source self.data.markledger(ledger, options=self.options) self.history.markledger(ledger, options=self.options) # Run repack self.repackdata(ledger, targetdata) self.repackhistory(ledger, targethistory) # Call cleanup on each source for source in ledger.sources: source.cleanup(ledger) def _chainorphans(self, ui, filename, nodes, orphans, deltabases): """Reorderes ``orphans`` into a single chain inside ``nodes`` and ``deltabases``. We often have orphan entries (nodes without a base that aren't referenced by other nodes -- i.e., part of a chain) due to gaps in history. Rather than store them as individual fulltexts, we prefer to insert them as one chain sorted by size. """ if not orphans: return nodes def getsize(node, default=0): meta = self.data.getmeta(filename, node) if constants.METAKEYSIZE in meta: return meta[constants.METAKEYSIZE] else: return default # Sort orphans by size; biggest first is preferred, since it's more # likely to be the newest version assuming files grow over time. # (Sort by node first to ensure the sort is stable.) orphans = sorted(orphans) orphans = list(sorted(orphans, key=getsize, reverse=True)) if ui.debugflag: ui.debug( b"%s: orphan chain: %s\n" % (filename, b", ".join([short(s) for s in orphans])) ) # Create one contiguous chain and reassign deltabases. for i, node in enumerate(orphans): if i == 0: deltabases[node] = (self.repo.nullid, 0) else: parent = orphans[i - 1] deltabases[node] = (parent, deltabases[parent][1] + 1) nodes = [n for n in nodes if n not in orphans] nodes += orphans return nodes def repackdata(self, ledger, target): ui = self.repo.ui maxchainlen = ui.configint(b'packs', b'maxchainlen', 1000) byfile = {} for entry in ledger.entries.values(): if entry.datasource: byfile.setdefault(entry.filename, {})[entry.node] = entry count = 0 repackprogress = ui.makeprogress( _(b"repacking data"), unit=self.unit, total=len(byfile) ) for filename, entries in sorted(byfile.items()): repackprogress.update(count) ancestors = {} nodes = list(node for node in entries) nohistory = [] buildprogress = ui.makeprogress( _(b"building history"), unit=b'nodes', total=len(nodes) ) for i, node in enumerate(nodes): if node in ancestors: continue buildprogress.update(i) try: ancestors.update( self.fullhistory.getancestors( filename, node, known=ancestors ) ) except KeyError: # Since we're packing data entries, we may not have the # corresponding history entries for them. It's not a big # deal, but the entries won't be delta'd perfectly. nohistory.append(node) buildprogress.complete() # Order the nodes children first, so we can produce reverse deltas orderednodes = list(reversed(self._toposort(ancestors))) if len(nohistory) > 0: ui.debug( b'repackdata: %d nodes without history\n' % len(nohistory) ) orderednodes.extend(sorted(nohistory)) # Filter orderednodes to just the nodes we want to serialize (it # currently also has the edge nodes' ancestors). orderednodes = list( filter(lambda node: node in nodes, orderednodes) ) # Garbage collect old nodes: if self.garbagecollect: neworderednodes = [] for node in orderednodes: # If the node is old and is not in the keepset, we skip it, # and mark as garbage collected if (filename, node) not in self.keepkeys and self.isold( self.repo, filename, node ): entries[node].gced = True continue neworderednodes.append(node) orderednodes = neworderednodes # Compute delta bases for nodes: deltabases = {} nobase = set() referenced = set() nodes = set(nodes) processprogress = ui.makeprogress( _(b"processing nodes"), unit=b'nodes', total=len(orderednodes) ) for i, node in enumerate(orderednodes): processprogress.update(i) # Find delta base # TODO: allow delta'ing against most recent descendant instead # of immediate child deltatuple = deltabases.get(node, None) if deltatuple is None: deltabase, chainlen = self.repo.nullid, 0 deltabases[node] = (self.repo.nullid, 0) nobase.add(node) else: deltabase, chainlen = deltatuple referenced.add(deltabase) # Use available ancestor information to inform our delta choices ancestorinfo = ancestors.get(node) if ancestorinfo: p1, p2, linknode, copyfrom = ancestorinfo # The presence of copyfrom means we're at a point where the # file was copied from elsewhere. So don't attempt to do any # deltas with the other file. if copyfrom: p1 = self.repo.nullid if chainlen < maxchainlen: # Record this child as the delta base for its parents. # This may be non optimal, since the parents may have # many children, and this will only choose the last one. # TODO: record all children and try all deltas to find # best if p1 != self.repo.nullid: deltabases[p1] = (node, chainlen + 1) if p2 != self.repo.nullid: deltabases[p2] = (node, chainlen + 1) # experimental config: repack.chainorphansbysize if ui.configbool(b'repack', b'chainorphansbysize'): orphans = nobase - referenced orderednodes = self._chainorphans( ui, filename, orderednodes, orphans, deltabases ) # Compute deltas and write to the pack for i, node in enumerate(orderednodes): deltabase, chainlen = deltabases[node] # Compute delta # TODO: Optimize the deltachain fetching. Since we're # iterating over the different version of the file, we may # be fetching the same deltachain over and over again. if deltabase != self.repo.nullid: deltaentry = self.data.getdelta(filename, node) delta, deltabasename, origdeltabase, meta = deltaentry size = meta.get(constants.METAKEYSIZE) if ( deltabasename != filename or origdeltabase != deltabase or size is None ): deltabasetext = self.data.get(filename, deltabase) original = self.data.get(filename, node) size = len(original) delta = mdiff.textdiff(deltabasetext, original) else: delta = self.data.get(filename, node) size = len(delta) meta = self.data.getmeta(filename, node) # TODO: don't use the delta if it's larger than the fulltext if constants.METAKEYSIZE not in meta: meta[constants.METAKEYSIZE] = size target.add(filename, node, deltabase, delta, meta) entries[node].datarepacked = True processprogress.complete() count += 1 repackprogress.complete() target.close(ledger=ledger) def repackhistory(self, ledger, target): ui = self.repo.ui byfile = {} for entry in ledger.entries.values(): if entry.historysource: byfile.setdefault(entry.filename, {})[entry.node] = entry progress = ui.makeprogress( _(b"repacking history"), unit=self.unit, total=len(byfile) ) for filename, entries in sorted(byfile.items()): ancestors = {} nodes = list(node for node in entries) for node in nodes: if node in ancestors: continue ancestors.update( self.history.getancestors(filename, node, known=ancestors) ) # Order the nodes children first orderednodes = reversed(self._toposort(ancestors)) # Write to the pack dontprocess = set() for node in orderednodes: p1, p2, linknode, copyfrom = ancestors[node] # If the node is marked dontprocess, but it's also in the # explicit entries set, that means the node exists both in this # file and in another file that was copied to this file. # Usually this happens if the file was copied to another file, # then the copy was deleted, then reintroduced without copy # metadata. The original add and the new add have the same hash # since the content is identical and the parents are null. if node in dontprocess and node not in entries: # If copyfrom == filename, it means the copy history # went to come other file, then came back to this one, so we # should continue processing it. if p1 != self.repo.nullid and copyfrom != filename: dontprocess.add(p1) if p2 != self.repo.nullid: dontprocess.add(p2) continue if copyfrom: dontprocess.add(p1) target.add(filename, node, p1, p2, linknode, copyfrom) if node in entries: entries[node].historyrepacked = True progress.increment() progress.complete() target.close(ledger=ledger) def _toposort(self, ancestors): def parentfunc(node): p1, p2, linknode, copyfrom = ancestors[node] parents = [] if p1 != self.repo.nullid: parents.append(p1) if p2 != self.repo.nullid: parents.append(p2) return parents sortednodes = shallowutil.sortnodes(ancestors.keys(), parentfunc) return sortednodes class repackledger: """Storage for all the bookkeeping that happens during a repack. It contains the list of revisions being repacked, what happened to each revision, and which source store contained which revision originally (for later cleanup). """ def __init__(self): self.entries = {} self.sources = {} self.created = set() def markdataentry(self, source, filename, node): """Mark the given filename+node revision as having a data rev in the given source. """ entry = self._getorcreateentry(filename, node) entry.datasource = True entries = self.sources.get(source) if not entries: entries = set() self.sources[source] = entries entries.add(entry) def markhistoryentry(self, source, filename, node): """Mark the given filename+node revision as having a history rev in the given source. """ entry = self._getorcreateentry(filename, node) entry.historysource = True entries = self.sources.get(source) if not entries: entries = set() self.sources[source] = entries entries.add(entry) def _getorcreateentry(self, filename, node): key = (filename, node) value = self.entries.get(key) if not value: value = repackentry(filename, node) self.entries[key] = value return value def addcreated(self, value): self.created.add(value) class repackentry: """Simple class representing a single revision entry in the repackledger.""" __slots__ = ( 'filename', 'node', 'datasource', 'historysource', 'datarepacked', 'historyrepacked', 'gced', ) def __init__(self, filename, node): self.filename = filename self.node = node # If the revision has a data entry in the source self.datasource = False # If the revision has a history entry in the source self.historysource = False # If the revision's data entry was repacked into the repack target self.datarepacked = False # If the revision's history entry was repacked into the repack target self.historyrepacked = False # If garbage collected self.gced = False def repacklockvfs(repo): if util.safehasattr(repo, 'name'): # Lock in the shared cache so repacks across multiple copies of the same # repo are coordinated. sharedcachepath = shallowutil.getcachepackpath( repo, constants.FILEPACK_CATEGORY ) return vfs.vfs(sharedcachepath) else: return repo.svfs