annotate hgext/largefiles/remotestore.py @ 17127:9e1616307c4c

largefiles: batch statlfile requests when pushing a largefiles repo (issue3386) This implements a part of issue 3386. It batches the request for the status of all largefiles in the revisions that are about to be pushed into a single request, instead of doing N separate requests. In a real world test case, this change was verified to save 1,116 round-trips to the server. It only requires a client-side change; it is backwards-compatible with an older version of the server.
author Na'Tosha Bard <natosha@unity3d.com>
date Sun, 24 Jun 2012 20:36:22 +0200
parents 67d010779907
children e95ec38f86b0
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
1 # Copyright 2010-2011 Fog Creek Software
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
2 # Copyright 2010-2011 Unity Technologies
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
3 #
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
4 # This software may be used and distributed according to the terms of the
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
5 # GNU General Public License version 2 or any later version.
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
6
15252
6e809bb4f969 largefiles: improve comments, internal docstrings
Greg Ward <greg@gerg.ca>
parents: 15188
diff changeset
7 '''remote largefile store; the base class for servestore'''
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
8
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
9 import urllib2
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
10
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
11 from mercurial import util
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
12 from mercurial.i18n import _
17127
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
13 from mercurial.wireproto import remotebatch
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
14
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
15 import lfutil
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
16 import basestore
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
17
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
18 class remotestore(basestore.basestore):
15252
6e809bb4f969 largefiles: improve comments, internal docstrings
Greg Ward <greg@gerg.ca>
parents: 15188
diff changeset
19 '''a largefile store accessed over a network'''
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
20 def __init__(self, ui, repo, url):
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
21 super(remotestore, self).__init__(ui, repo, url)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
22
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
23 def put(self, source, hash):
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
24 if self.sendfile(source, hash):
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
25 raise util.Abort(
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
26 _('remotestore: could not put %s to remote store %s')
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
27 % (source, self.url))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
28 self.ui.debug(
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
29 _('remotestore: put %s to remote store %s') % (source, self.url))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
30
17127
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
31 def exists(self, hashes):
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
32 return self._verify(hashes)
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
33
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
34 def sendfile(self, filename, hash):
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
35 self.ui.debug('remotestore: sendfile(%s, %s)\n' % (filename, hash))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
36 fd = None
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
37 try:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
38 try:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
39 fd = lfutil.httpsendfile(self.ui, filename)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
40 except IOError, e:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
41 raise util.Abort(
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
42 _('remotestore: could not open file %s: %s')
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
43 % (filename, str(e)))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
44 return self._put(hash, fd)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
45 finally:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
46 if fd:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
47 fd.close()
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
48
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
49 def _getfile(self, tmpfile, filename, hash):
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
50 # quit if the largefile isn't there
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
51 stat = self._stat(hash)
15253
67d010779907 largefiles: improve error reporting
Greg Ward <greg@gerg.ca>
parents: 15252
diff changeset
52 if stat == 1:
67d010779907 largefiles: improve error reporting
Greg Ward <greg@gerg.ca>
parents: 15252
diff changeset
53 raise util.Abort(_('remotestore: largefile %s is invalid') % hash)
67d010779907 largefiles: improve error reporting
Greg Ward <greg@gerg.ca>
parents: 15252
diff changeset
54 elif stat == 2:
67d010779907 largefiles: improve error reporting
Greg Ward <greg@gerg.ca>
parents: 15252
diff changeset
55 raise util.Abort(_('remotestore: largefile %s is missing') % hash)
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
56
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
57 try:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
58 length, infile = self._get(hash)
15188
8e115063950d largefiles: don't break existing tests (syntax error, bad imports)
Greg Ward <greg@gerg.ca>
parents: 15168
diff changeset
59 except urllib2.HTTPError, e:
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
60 # 401s get converted to util.Aborts; everything else is fine being
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
61 # turned into a StoreError
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
62 raise basestore.StoreError(filename, hash, self.url, str(e))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
63 except urllib2.URLError, e:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
64 # This usually indicates a connection problem, so don't
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
65 # keep trying with the other files... they will probably
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
66 # all fail too.
15253
67d010779907 largefiles: improve error reporting
Greg Ward <greg@gerg.ca>
parents: 15252
diff changeset
67 raise util.Abort('%s: %s' % (self.url, e.reason))
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
68 except IOError, e:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
69 raise basestore.StoreError(filename, hash, self.url, str(e))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
70
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
71 # Mercurial does not close its SSH connections after writing a stream
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
72 if length is not None:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
73 infile = lfutil.limitreader(infile, length)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
74 return lfutil.copyandhash(lfutil.blockstream(infile), tmpfile)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
75
17127
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
76 def _verify(self, hashes):
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
77 return self._stat(hashes)
15168
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
78
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
79 def _verifyfile(self, cctx, cset, contents, standin, verified):
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
80 filename = lfutil.splitstandin(standin)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
81 if not filename:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
82 return False
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
83 fctx = cctx[standin]
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
84 key = (filename, fctx.filenode())
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
85 if key in verified:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
86 return False
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
87
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
88 verified.add(key)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
89
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
90 stat = self._stat(hash)
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
91 if not stat:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
92 return False
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
93 elif stat == 1:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
94 self.ui.warn(
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
95 _('changeset %s: %s: contents differ\n')
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
96 % (cset, filename))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
97 return True # failed
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
98 elif stat == 2:
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
99 self.ui.warn(
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
100 _('changeset %s: %s missing\n')
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
101 % (cset, filename))
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
102 return True # failed
cfccd3bee7b3 hgext: add largefiles extension
various
parents:
diff changeset
103 else:
15253
67d010779907 largefiles: improve error reporting
Greg Ward <greg@gerg.ca>
parents: 15252
diff changeset
104 raise RuntimeError('verify failed: unexpected response from '
67d010779907 largefiles: improve error reporting
Greg Ward <greg@gerg.ca>
parents: 15252
diff changeset
105 'statlfile (%r)' % stat)
17127
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
106
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
107 def batch(self):
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
108 '''Support for remote batching.'''
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
109 return remotebatch(self)
9e1616307c4c largefiles: batch statlfile requests when pushing a largefiles repo (issue3386)
Na'Tosha Bard <natosha@unity3d.com>
parents: 15253
diff changeset
110