view tests/test-archive.t @ 26623:5a95fe44121d

clonebundles: support for seeding clones from pre-generated bundles

Cloning can be an expensive operation for servers because the server
generates a bundle from existing repository data at request time. For
a large repository like mozilla-central, this consumes 4+ minutes
of CPU time on the server. It also results in significant network
utilization. Multiplied by hundreds or even thousands of clients and
the ensuing load can result in difficulties scaling the Mercurial server.

Despite generation of bundles being deterministic until the next
changeset is added, the generation of bundles to service a clone request
is not cached. Each clone thus performs redundant work. This is
wasteful.

This patch introduces the "clonebundles" extension and related
client-side functionality to help alleviate this deficiency. The
client-side feature is behind an experimental flag and is not enabled by
default.

It works as follows:

1) Server operator generates a bundle and makes it available on a
   server (likely HTTP).
2) Server operator defines the URL of a bundle file in a
   .hg/clonebundles.manifest file.
3) Client `hg clone`ing sees the server is advertising bundle URLs.
4) Client fetches and applies the advertised bundle.
5) Client performs equivalent of `hg pull` to fetch changes made since
   the bundle was created.

Essentially, the server performs the expensive work of generating a
bundle once and all subsequent clones fetch a static file from
somewhere. Scaling static file serving is a much more manageable
problem than scaling a Python application like Mercurial. Assuming your
repository grows less than 1% per day, the end result is 99+% of CPU
and network load from clones is eliminated, allowing Mercurial servers
to scale more easily. Serving static files also means data can be
transferred to clients as fast as they can consume it, rather than as
fast as servers can generate it. This makes clones faster.

Mozilla has implemented similar functionality of this patch on
hg.mozilla.org using a custom extension. We are hosting bundle files in
Amazon S3 and CloudFront (a CDN) and have successfully offloaded
>1 TB/day in data transfer from hg.mozilla.org, freeing up significant
bandwidth and CPU resources. The positive impact has been stellar and
I believe it has proved its value to be included in Mercurial core. I
feel it is important for the client-side support to be enabled in core
by default because it means that clients will get faster, more reliable
clones and will enable server operators to reduce load without
requiring any client-side configuration changes (assuming clients are
up to date, of course).

The scope of this feature is narrowly and specifically tailored to
cloning, despite "serve pulls from pre-generated bundles" being a valid
and useful feature. I would eventually like for Mercurial servers to
support transferring *all* repository data via statically hosted files.
You could imagine a server that siphons all pushed data to bundle files
and instructs clients to apply a stream of bundles to reconstruct all
repository data. This feature, while useful and powerful, is
significantly more work to implement because it requires the server
component have awareness of discovery and a mapping of which changesets
are in which files. Full, clone bundles, by contrast, are much simpler.

The wire protocol command is named "clonebundles" instead of something
more generic like "staticbundles" to leave the door open for a new, more
powerful and more generic server-side component with minimal backwards
compatibility implications. The name "bundleclone" is used by Mozilla's
extension and would cause problems since there are subtle differences
in Mozilla's extension.

Mozilla's experience with this idea has taught us that some form of
"content negotiation" is required. Not all clients will support all
bundle formats or even URLs (advanced TLS requirements, etc). To ensure
the highest uptake possible, a server needs to advertise multiple
versions of bundles and clients need to be able to choose the most
appropriate from that list one. The "attributes" in each
server-advertised entry facilitate this filtering and sorting. Their
use will become apparent in subsequent patches.

Initial inspiration and credit for the idea of cloning from static files
belongs to Augie Fackler and his "lookaside clone" extension proof of
concept.

#require serve

  $ hg init test
  $ cd test
  $ echo foo>foo
  $ hg commit -Am 1 -d '1 0'
  adding foo
  $ echo bar>bar
  $ hg commit -Am 2 -d '2 0'
  adding bar
  $ mkdir baz
  $ echo bletch>baz/bletch
  $ hg commit -Am 3 -d '1000000000 0'
  adding baz/bletch
  $ hg init subrepo
  $ touch subrepo/sub
  $ hg -q -R subrepo ci -Am "init subrepo"
  $ echo "subrepo = subrepo" > .hgsub
  $ hg add .hgsub
  $ hg ci -m "add subrepo"
  $ echo "[web]" >> .hg/hgrc
  $ echo "name = test-archive" >> .hg/hgrc
  $ echo "archivesubrepos = True" >> .hg/hgrc
  $ cp .hg/hgrc .hg/hgrc-base
  > test_archtype() {
  >     echo "allow_archive = $1" >> .hg/hgrc
  >     hg serve -p $HGPORT -d --pid-file=hg.pid -E errors.log
  >     cat hg.pid >> $DAEMON_PIDS
  >     echo % $1 allowed should give 200
  >     get-with-headers.py localhost:$HGPORT "archive/tip.$2" | head -n 1
  >     echo % $3 and $4 disallowed should both give 403
  >     get-with-headers.py localhost:$HGPORT "archive/tip.$3" | head -n 1
  >     get-with-headers.py localhost:$HGPORT "archive/tip.$4" | head -n 1
  >     killdaemons.py
  >     cat errors.log
  >     cp .hg/hgrc-base .hg/hgrc
  > }

check http return codes

  $ test_archtype gz tar.gz tar.bz2 zip
  % gz allowed should give 200
  200 Script output follows
  % tar.bz2 and zip disallowed should both give 403
  403 Archive type not allowed: bz2
  403 Archive type not allowed: zip
  $ test_archtype bz2 tar.bz2 zip tar.gz
  % bz2 allowed should give 200
  200 Script output follows
  % zip and tar.gz disallowed should both give 403
  403 Archive type not allowed: zip
  403 Archive type not allowed: gz
  $ test_archtype zip zip tar.gz tar.bz2
  % zip allowed should give 200
  200 Script output follows
  % tar.gz and tar.bz2 disallowed should both give 403
  403 Archive type not allowed: gz
  403 Archive type not allowed: bz2

  $ echo "allow_archive = gz bz2 zip" >> .hg/hgrc
  $ hg serve -p $HGPORT -d --pid-file=hg.pid -E errors.log
  $ cat hg.pid >> $DAEMON_PIDS

invalid arch type should give 404

  $ get-with-headers.py localhost:$HGPORT "archive/tip.invalid" | head -n 1
  404 Unsupported archive type: None

  $ TIP=`hg id -v | cut -f1 -d' '`
  $ QTIP=`hg id -q`
  $ cat > getarchive.py <<EOF
  > import os, sys, urllib2
  > try:
  >     # Set stdout to binary mode for win32 platforms
  >     import msvcrt
  >     msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
  > except ImportError:
  >     pass
  > if len(sys.argv) <= 3:
  >     node, archive = sys.argv[1:]
  >     requeststr = 'cmd=archive;node=%s;type=%s' % (node, archive)
  > else:
  >     node, archive, file = sys.argv[1:]
  >     requeststr = 'cmd=archive;node=%s;type=%s;file=%s' % (node, archive, file)
  > try:
  >     f = urllib2.urlopen('http://127.0.0.1:%s/?%s'
  >                     % (os.environ['HGPORT'], requeststr))
  >     sys.stdout.write(f.read())
  > except urllib2.HTTPError, e:
  >     sys.stderr.write(str(e) + '\n')
  > EOF
  $ python getarchive.py "$TIP" gz | gunzip | tar tf - 2>/dev/null
  test-archive-1701ef1f1510/.hg_archival.txt
  test-archive-1701ef1f1510/.hgsub
  test-archive-1701ef1f1510/.hgsubstate
  test-archive-1701ef1f1510/bar
  test-archive-1701ef1f1510/baz/bletch
  test-archive-1701ef1f1510/foo
  test-archive-1701ef1f1510/subrepo/sub
  $ python getarchive.py "$TIP" bz2 | bunzip2 | tar tf - 2>/dev/null
  test-archive-1701ef1f1510/.hg_archival.txt
  test-archive-1701ef1f1510/.hgsub
  test-archive-1701ef1f1510/.hgsubstate
  test-archive-1701ef1f1510/bar
  test-archive-1701ef1f1510/baz/bletch
  test-archive-1701ef1f1510/foo
  test-archive-1701ef1f1510/subrepo/sub
  $ python getarchive.py "$TIP" zip > archive.zip
  $ unzip -t archive.zip
  Archive:  archive.zip
      testing: test-archive-1701ef1f1510/.hg_archival.txt   OK
      testing: test-archive-1701ef1f1510/.hgsub   OK
      testing: test-archive-1701ef1f1510/.hgsubstate   OK
      testing: test-archive-1701ef1f1510/bar   OK
      testing: test-archive-1701ef1f1510/baz/bletch   OK
      testing: test-archive-1701ef1f1510/foo   OK
      testing: test-archive-1701ef1f1510/subrepo/sub   OK
  No errors detected in compressed data of archive.zip.

test that we can download single directories and files

  $ python getarchive.py "$TIP" gz baz | gunzip | tar tf - 2>/dev/null
  test-archive-1701ef1f1510/baz/bletch
  $ python getarchive.py "$TIP" gz foo | gunzip | tar tf - 2>/dev/null
  test-archive-1701ef1f1510/foo

test that we detect file patterns that match no files

  $ python getarchive.py "$TIP" gz foobar
  HTTP Error 404: file(s) not found: foobar

test that we reject unsafe patterns

  $ python getarchive.py "$TIP" gz relre:baz
  HTTP Error 404: file(s) not found: relre:baz

  $ killdaemons.py

  $ hg archive -t tar test.tar
  $ tar tf test.tar
  test/.hg_archival.txt
  test/.hgsub
  test/.hgsubstate
  test/bar
  test/baz/bletch
  test/foo

  $ hg archive --debug -t tbz2 -X baz test.tar.bz2 --config progress.debug=true
  archiving: 0/4 files (0.00%)
  archiving: .hgsub 1/4 files (25.00%)
  archiving: .hgsubstate 2/4 files (50.00%)
  archiving: bar 3/4 files (75.00%)
  archiving: foo 4/4 files (100.00%)
  $ bunzip2 -dc test.tar.bz2 | tar tf - 2>/dev/null
  test/.hg_archival.txt
  test/.hgsub
  test/.hgsubstate
  test/bar
  test/foo

  $ hg archive -t tgz -p %b-%h test-%h.tar.gz
  $ gzip -dc test-$QTIP.tar.gz | tar tf - 2>/dev/null
  test-1701ef1f1510/.hg_archival.txt
  test-1701ef1f1510/.hgsub
  test-1701ef1f1510/.hgsubstate
  test-1701ef1f1510/bar
  test-1701ef1f1510/baz/bletch
  test-1701ef1f1510/foo

  $ hg archive autodetected_test.tar
  $ tar tf autodetected_test.tar
  autodetected_test/.hg_archival.txt
  autodetected_test/.hgsub
  autodetected_test/.hgsubstate
  autodetected_test/bar
  autodetected_test/baz/bletch
  autodetected_test/foo

The '-t' should override autodetection

  $ hg archive -t tar autodetect_override_test.zip
  $ tar tf autodetect_override_test.zip
  autodetect_override_test.zip/.hg_archival.txt
  autodetect_override_test.zip/.hgsub
  autodetect_override_test.zip/.hgsubstate
  autodetect_override_test.zip/bar
  autodetect_override_test.zip/baz/bletch
  autodetect_override_test.zip/foo

  $ for ext in tar tar.gz tgz tar.bz2 tbz2 zip; do
  >     hg archive auto_test.$ext
  >     if [ -d auto_test.$ext ]; then
  >         echo "extension $ext was not autodetected."
  >     fi
  > done

  $ cat > md5comp.py <<EOF
  > try:
  >     from hashlib import md5
  > except ImportError:
  >     from md5 import md5
  > import sys
  > f1, f2 = sys.argv[1:3]
  > h1 = md5(file(f1, 'rb').read()).hexdigest()
  > h2 = md5(file(f2, 'rb').read()).hexdigest()
  > print h1 == h2 or "md5 differ: " + repr((h1, h2))
  > EOF

archive name is stored in the archive, so create similar archives and
rename them afterwards.

  $ hg archive -t tgz tip.tar.gz
  $ mv tip.tar.gz tip1.tar.gz
  $ sleep 1
  $ hg archive -t tgz tip.tar.gz
  $ mv tip.tar.gz tip2.tar.gz
  $ python md5comp.py tip1.tar.gz tip2.tar.gz
  True

  $ hg archive -t zip -p /illegal test.zip
  abort: archive prefix contains illegal components
  [255]
  $ hg archive -t zip -p very/../bad test.zip

  $ hg archive --config ui.archivemeta=false -t zip -r 2 test.zip
  $ unzip -t test.zip
  Archive:  test.zip
      testing: test/bar                 OK
      testing: test/baz/bletch          OK
      testing: test/foo                 OK
  No errors detected in compressed data of test.zip.

  $ hg archive -t tar - | tar tf - 2>/dev/null
  test-1701ef1f1510/.hg_archival.txt
  test-1701ef1f1510/.hgsub
  test-1701ef1f1510/.hgsubstate
  test-1701ef1f1510/bar
  test-1701ef1f1510/baz/bletch
  test-1701ef1f1510/foo

  $ hg archive -r 0 -t tar rev-%r.tar
  $ [ -f rev-0.tar ]

test .hg_archival.txt

  $ hg archive ../test-tags
  $ cat ../test-tags/.hg_archival.txt
  repo: daa7f7c60e0a224faa4ff77ca41b2760562af264
  node: 1701ef1f151069b8747038e93b5186bb43a47504
  branch: default
  latesttag: null
  latesttagdistance: 4
  changessincelatesttag: 4
  $ hg tag -r 2 mytag
  $ hg tag -r 2 anothertag
  $ hg archive -r 2 ../test-lasttag
  $ cat ../test-lasttag/.hg_archival.txt
  repo: daa7f7c60e0a224faa4ff77ca41b2760562af264
  node: 2c0277f05ed49d1c8328fb9ba92fba7a5ebcb33e
  branch: default
  tag: anothertag
  tag: mytag

  $ hg archive -t bogus test.bogus
  abort: unknown archive type 'bogus'
  [255]

enable progress extension:

  $ cp $HGRCPATH $HGRCPATH.no-progress
  $ cat >> $HGRCPATH <<EOF
  > [extensions]
  > progress =
  > [progress]
  > assume-tty = 1
  > format = topic bar number
  > delay = 0
  > refresh = 0
  > width = 60
  > EOF

  $ hg archive ../with-progress
  \r (no-eol) (esc)
  archiving [                                           ] 0/6\r (no-eol) (esc)
  archiving [======>                                    ] 1/6\r (no-eol) (esc)
  archiving [=============>                             ] 2/6\r (no-eol) (esc)
  archiving [====================>                      ] 3/6\r (no-eol) (esc)
  archiving [===========================>               ] 4/6\r (no-eol) (esc)
  archiving [==================================>        ] 5/6\r (no-eol) (esc)
  archiving [==========================================>] 6/6\r (no-eol) (esc)
                                                              \r (no-eol) (esc)

cleanup after progress extension test:

  $ cp $HGRCPATH.no-progress $HGRCPATH

server errors

  $ cat errors.log

empty repo

  $ hg init ../empty
  $ cd ../empty
  $ hg archive ../test-empty
  abort: no working directory: please specify a revision
  [255]

old file -- date clamped to 1980

  $ touch -t 197501010000 old
  $ hg add old
  $ hg commit -m old
  $ hg archive ../old.zip
  $ unzip -l ../old.zip
  Archive:  ../old.zip
  \s*Length.* (re)
  *-----* (glob)
  *172*80*00:00*old/.hg_archival.txt (glob)
  *0*80*00:00*old/old (glob)
  *-----* (glob)
  \s*172\s+2 files (re)

show an error when a provided pattern matches no files

  $ hg archive -I file_that_does_not_exist.foo ../empty.zip
  abort: no files match the archive pattern
  [255]

  $ hg archive -X * ../empty.zip
  abort: no files match the archive pattern
  [255]

  $ cd ..

issue3600: check whether "hg archive" can create archive files which
are extracted with expected timestamp, even though TZ is not
configured as GMT.

  $ mkdir issue3600
  $ cd issue3600

  $ hg init repo
  $ echo a > repo/a
  $ hg -R repo add repo/a
  $ hg -R repo commit -m '#0' -d '456789012 21600'
  $ cat > show_mtime.py <<EOF
  > import sys, os
  > print int(os.stat(sys.argv[1]).st_mtime)
  > EOF

  $ hg -R repo archive --prefix tar-extracted archive.tar
  $ (TZ=UTC-3; export TZ; tar xf archive.tar)
  $ python show_mtime.py tar-extracted/a
  456789012

  $ hg -R repo archive --prefix zip-extracted archive.zip
  $ (TZ=UTC-3; export TZ; unzip -q archive.zip)
  $ python show_mtime.py zip-extracted/a
  456789012

  $ cd ..