view hgext/remotefilelog/metadatastore.py @ 42354:dbd0fcca6dfc

discovery: slowly increase sampling size

Some pathological discovery runs can requires many roundtrip. When this happens
things can get very slow.

To make the algorithm more resilience again such pathological case. We slowly
increase the sample size with each roundtrip (+5%). This will have a negligible
impact on "normal" discovery with few roundtrips, but a large positive impact of
case with many roundtrips. Asking more question per roundtrip helps to reduce
the undecided set faster. Instead of reducing the undecided set a linear speed
(in the worst case), we reduce it as a guaranteed (small) exponential rate. The
data below show this slow ramp up in sample size:

round trip    |    1 |     5 |    10 |    20 |     50 |     100 |       130 |
sample size   |  200 |   254 |   321 |   517 |  2 199 |  25 123 |   108 549 |
covered nodes |  200 | 1 357 | 2 821 | 7 031 | 42 658 | 524 530 | 2 276 755 |

To be a bit more concrete, lets take a very pathological case as an example. We
are doing discovery from a copy of Mozilla-try to a more recent version of
mozilla-unified. Mozilla-unified heads are unknown to the mozilla-try repo and
there are over 1 million "missing" changesets. (the discovery is "local" to
avoid network interference)

Without this change, the discovery:
- last 1858 seconds (31 minutes),
- does 1700 round trip,
- asking about 340 000 nodes.

With this change, the discovery:
- last 218 seconds (3 minutes, 38 seconds a -88% improvement),
- does 94 round trip (-94%),
- asking about 344 211 nodes (+1%).

Of course, this is an extreme case (and 3 minutes is still slow). However this
give a good example of how this sample size increase act as a safety net
catching any bad situations.

We could image a steeper increase than 5%. For example 10% would give the
following number:

round trip    |    1 |     5 |    10 |     20 |     50  |        75 |        100 |
sample size   |  200 |   321 |   514 |  1 326 |  23 060 |   249 812 |  2 706 594 |
covered nodes |  200 | 1 541 | 3 690 | 12 671 | 251 871 | 2 746 254 | 29 770 966 |

In parallel, it is useful to understand these pathological cases and improve
them. However the current change provides a general purpose safety net to smooth
the impact of pathological cases.

To avoid issue with older http server, the increase in sample size only occurs
if the protocol has not limit on command argument size.

from __future__ import absolute_import

from mercurial.node import hex, nullid
from . import (
    basestore,
    shallowutil,
)

class unionmetadatastore(basestore.baseunionstore):
    def __init__(self, *args, **kwargs):
        super(unionmetadatastore, self).__init__(*args, **kwargs)

        self.stores = args
        self.writestore = kwargs.get(r'writestore')

        # If allowincomplete==True then the union store can return partial
        # ancestor lists, otherwise it will throw a KeyError if a full
        # history can't be found.
        self.allowincomplete = kwargs.get(r'allowincomplete', False)

    def getancestors(self, name, node, known=None):
        """Returns as many ancestors as we're aware of.

        return value: {
           node: (p1, p2, linknode, copyfrom),
           ...
        }
        """
        if known is None:
            known = set()
        if node in known:
            return []

        ancestors = {}
        def traverse(curname, curnode):
            # TODO: this algorithm has the potential to traverse parts of
            # history twice. Ex: with A->B->C->F and A->B->D->F, both D and C
            # may be queued as missing, then B and A are traversed for both.
            queue = [(curname, curnode)]
            missing = []
            seen = set()
            while queue:
                name, node = queue.pop()
                if (name, node) in seen:
                    continue
                seen.add((name, node))
                value = ancestors.get(node)
                if not value:
                    missing.append((name, node))
                    continue
                p1, p2, linknode, copyfrom = value
                if p1 != nullid and p1 not in known:
                    queue.append((copyfrom or curname, p1))
                if p2 != nullid and p2 not in known:
                    queue.append((curname, p2))
            return missing

        missing = [(name, node)]
        while missing:
            curname, curnode = missing.pop()
            try:
                ancestors.update(self._getpartialancestors(curname, curnode,
                                                           known=known))
                newmissing = traverse(curname, curnode)
                missing.extend(newmissing)
            except KeyError:
                # If we allow incomplete histories, don't throw.
                if not self.allowincomplete:
                    raise
                # If the requested name+node doesn't exist, always throw.
                if (curname, curnode) == (name, node):
                    raise

        # TODO: ancestors should probably be (name, node) -> (value)
        return ancestors

    @basestore.baseunionstore.retriable
    def _getpartialancestors(self, name, node, known=None):
        for store in self.stores:
            try:
                return store.getancestors(name, node, known=known)
            except KeyError:
                pass

        raise KeyError((name, hex(node)))

    @basestore.baseunionstore.retriable
    def getnodeinfo(self, name, node):
        for store in self.stores:
            try:
                return store.getnodeinfo(name, node)
            except KeyError:
                pass

        raise KeyError((name, hex(node)))

    def add(self, name, node, data):
        raise RuntimeError("cannot add content only to remotefilelog "
                           "contentstore")

    def getmissing(self, keys):
        missing = keys
        for store in self.stores:
            if missing:
                missing = store.getmissing(missing)
        return missing

    def markledger(self, ledger, options=None):
        for store in self.stores:
            store.markledger(ledger, options)

    def getmetrics(self):
        metrics = [s.getmetrics() for s in self.stores]
        return shallowutil.sumdicts(*metrics)

class remotefilelogmetadatastore(basestore.basestore):
    def getancestors(self, name, node, known=None):
        """Returns as many ancestors as we're aware of.

        return value: {
           node: (p1, p2, linknode, copyfrom),
           ...
        }
        """
        data = self._getdata(name, node)
        ancestors = shallowutil.ancestormap(data)
        return ancestors

    def getnodeinfo(self, name, node):
        return self.getancestors(name, node)[node]

    def add(self, name, node, parents, linknode):
        raise RuntimeError("cannot add metadata only to remotefilelog "
                           "metadatastore")

class remotemetadatastore(object):
    def __init__(self, ui, fileservice, shared):
        self._fileservice = fileservice
        self._shared = shared

    def getancestors(self, name, node, known=None):
        self._fileservice.prefetch([(name, hex(node))], force=True,
                                   fetchdata=False, fetchhistory=True)
        return self._shared.getancestors(name, node, known=known)

    def getnodeinfo(self, name, node):
        return self.getancestors(name, node)[node]

    def add(self, name, node, data):
        raise RuntimeError("cannot add to a remote store")

    def getmissing(self, keys):
        return keys

    def markledger(self, ledger, options=None):
        pass