##// END OF EJS Templates
upstream » mercurial-mirror
RSS

Clone from https://www.mercurial-scm.org/repo/hg-all

discovery: slowly increase sampling size...
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.
Gregory Szorc - - Load All Authors

File last commit:

r36469:1fa35ca3 default
r42546:dbd0fcca default
Show More
bundles.txt
93 lines | 3.2 KiB | text/plain | TextLexer
/ mercurial / help / internals / bundles.txt
History | Annotation | Raw |Copy content |Copy permalink
A bundle is a container for repository data.
Bundles are used as standalone files as well as the interchange format
over the wire protocol used when two Mercurial peers communicate with
each other.
Headers
=======
Bundles produced since Mercurial 0.7 (September 2005) have a 4 byte
header identifying the major bundle type. The header always begins with
``HG`` and the follow 2 bytes indicate the bundle type/version. Some
bundle types have additional data after this 4 byte header.
The following sections describe each bundle header/type.
HG10
----
``HG10`` headers indicate a *changegroup bundle*. This is the original
bundle format, so it is sometimes referred to as *bundle1*. It has been
present since version 0.7 (released September 2005).
This header is followed by 2 bytes indicating the compression algorithm
used for data that follows. All subsequent data following this
compression identifier is compressed according to the algorithm/method
specified.
Supported algorithms include the following.
``BZ``
*bzip2* compression.
Bzip2 compressors emit a leading ``BZ`` header. Mercurial uses this
leading ``BZ`` as part of the bundle header. Therefore consumers
of bzip2 bundles need to *seed* the bzip2 decompressor with ``BZ`` or
seek the input stream back to the beginning of the algorithm component
of the bundle header so that decompressor input is valid. This behavior
is unique among supported compression algorithms.
Supported since version 0.7 (released December 2006).
``GZ``
*zlib* compression.
Supported since version 0.9.2 (released December 2006).
``UN``
*Uncompressed* or no compression. Unmodified changegroup data follows.
Supported since version 0.9.2 (released December 2006).
3rd party extensions may implement their own compression. However, no
authority reserves values for their compression algorithm identifiers.
HG2X
----
``HG2X`` headers (where ``X`` is any value) denote a *bundle2* bundle.
Bundle2 bundles are a container format for various kinds of repository
data and capabilities, beyond changegroup data (which was the only data
supported by ``HG10`` bundles.
``HG20`` is currently the only defined bundle2 version.
The ``HG20`` format is documented at :hg:`help internals.bundle2`.
Initial ``HG20`` support was added in Mercurial 3.0 (released May
2014). However, bundle2 bundles were hidden behind an experimental flag
until version 3.5 (released August 2015), when they were enabled in the
wire protocol. Various commands (including ``hg bundle``) did not
support generating bundle2 files until Mercurial 3.6 (released November
2015).
HGS1
----
*Experimental*
A ``HGS1`` header indicates a *streaming clone bundle*. This is a bundle
that contains raw revlog data from a repository store. (Typically revlog
data is exchanged in the form of changegroups.)
The purpose of *streaming clone bundles* are to *clone* repository data
very efficiently.
The ``HGS1`` header is always followed by 2 bytes indicating a
compression algorithm of the data that follows. Only ``UN``
(uncompressed data) is currently allowed.
``HGS1UN`` support was added as an experimental feature in version 3.6
(released November 2015) as part of the initial offering of the *clone
bundles* feature.