##// END OF EJS Templates
clonebundles: fix typo s/comand/command/
Javi Merino -
r26884:762bf510 stable
parent child Browse files
Show More
@@ -1,254 +1,254 b''
1 1 # This software may be used and distributed according to the terms of the
2 2 # GNU General Public License version 2 or any later version.
3 3
4 4 """advertise pre-generated bundles to seed clones (experimental)
5 5
6 6 "clonebundles" is a server-side extension used to advertise the existence
7 7 of pre-generated, externally hosted bundle files to clients that are
8 8 cloning so that cloning can be faster, more reliable, and require less
9 9 resources on the server.
10 10
11 11 Cloning can be a CPU and I/O intensive operation on servers. Traditionally,
12 12 the server, in response to a client's request to clone, dynamically generates
13 13 a bundle containing the entire repository content and sends it to the client.
14 14 There is no caching on the server and the server will have to redundantly
15 15 generate the same outgoing bundle in response to each clone request. For
16 16 servers with large repositories or with high clone volume, the load from
17 17 clones can make scaling the server challenging and costly.
18 18
19 19 This extension provides server operators the ability to offload potentially
20 20 expensive clone load to an external service. Here's how it works.
21 21
22 22 1. A server operator establishes a mechanism for making bundle files available
23 23 on a hosting service where Mercurial clients can fetch them.
24 24 2. A manifest file listing available bundle URLs and some optional metadata
25 25 is added to the Mercurial repository on the server.
26 26 3. A client initiates a clone against a clone bundles aware server.
27 27 4. The client sees the server is advertising clone bundles and fetches the
28 28 manifest listing available bundles.
29 29 5. The client filters and sorts the available bundles based on what it
30 30 supports and prefers.
31 31 6. The client downloads and applies an available bundle from the
32 32 server-specified URL.
33 33 7. The client reconnects to the original server and performs the equivalent
34 34 of :hg:`pull` to retrieve all repository data not in the bundle. (The
35 35 repository could have been updated between when the bundle was created
36 36 and when the client started the clone.)
37 37
38 38 Instead of the server generating full repository bundles for every clone
39 39 request, it generates full bundles once and they are subsequently reused to
40 40 bootstrap new clones. The server may still transfer data at clone time.
41 41 However, this is only data that has been added/changed since the bundle was
42 42 created. For large, established repositories, this can reduce server load for
43 43 clones to less than 1% of original.
44 44
45 45 To work, this extension requires the following of server operators:
46 46
47 47 * Generating bundle files of repository content (typically periodically,
48 48 such as once per day).
49 49 * A file server that clients have network access to and that Python knows
50 50 how to talk to through its normal URL handling facility (typically a
51 51 HTTP server).
52 52 * A process for keeping the bundles manifest in sync with available bundle
53 53 files.
54 54
55 55 Strictly speaking, using a static file hosting server isn't required: a server
56 56 operator could use a dynamic service for retrieving bundle data. However,
57 57 static file hosting services are simple and scalable and should be sufficient
58 58 for most needs.
59 59
60 Bundle files can be generated with the :hg:`bundle` comand. Typically
60 Bundle files can be generated with the :hg:`bundle` command. Typically
61 61 :hg:`bundle --all` is used to produce a bundle of the entire repository.
62 62
63 63 :hg:`debugcreatestreamclonebundle` can be used to produce a special
64 64 *streaming clone bundle*. These are bundle files that are extremely efficient
65 65 to produce and consume (read: fast). However, they are larger than
66 66 traditional bundle formats and require that clients support the exact set
67 67 of repository data store formats in use by the repository that created them.
68 68 Typically, a newer server can serve data that is compatible with older clients.
69 69 However, *streaming clone bundles* don't have this guarantee. **Server
70 70 operators need to be aware that newer versions of Mercurial may produce
71 71 streaming clone bundles incompatible with older Mercurial versions.**
72 72
73 73 The list of requirements printed by :hg:`debugcreatestreamclonebundle` should
74 74 be specified in the ``requirements`` parameter of the *bundle specification
75 75 string* for the ``BUNDLESPEC`` manifest property described below. e.g.
76 76 ``BUNDLESPEC=none-packed1;requirements%3Drevlogv1``.
77 77
78 78 A server operator is responsible for creating a ``.hg/clonebundles.manifest``
79 79 file containing the list of available bundle files suitable for seeding
80 80 clones. If this file does not exist, the repository will not advertise the
81 81 existence of clone bundles when clients connect.
82 82
83 83 The manifest file contains a newline (\n) delimited list of entries.
84 84
85 85 Each line in this file defines an available bundle. Lines have the format:
86 86
87 87 <URL> [<key>=<value>[ <key>=<value>]]
88 88
89 89 That is, a URL followed by an optional, space-delimited list of key=value
90 90 pairs describing additional properties of this bundle. Both keys and values
91 91 are URI encoded.
92 92
93 93 Keys in UPPERCASE are reserved for use by Mercurial and are defined below.
94 94 All non-uppercase keys can be used by site installations. An example use
95 95 for custom properties is to use the *datacenter* attribute to define which
96 96 data center a file is hosted in. Clients could then prefer a server in the
97 97 data center closest to them.
98 98
99 99 The following reserved keys are currently defined:
100 100
101 101 BUNDLESPEC
102 102 A "bundle specification" string that describes the type of the bundle.
103 103
104 104 These are string values that are accepted by the "--type" argument of
105 105 :hg:`bundle`.
106 106
107 107 The values are parsed in strict mode, which means they must be of the
108 108 "<compression>-<type>" form. See
109 109 mercurial.exchange.parsebundlespec() for more details.
110 110
111 111 Clients will automatically filter out specifications that are unknown or
112 112 unsupported so they won't attempt to download something that likely won't
113 113 apply.
114 114
115 115 The actual value doesn't impact client behavior beyond filtering:
116 116 clients will still sniff the bundle type from the header of downloaded
117 117 files.
118 118
119 119 **Use of this key is highly recommended**, as it allows clients to
120 120 easily skip unsupported bundles.
121 121
122 122 REQUIRESNI
123 123 Whether Server Name Indication (SNI) is required to connect to the URL.
124 124 SNI allows servers to use multiple certificates on the same IP. It is
125 125 somewhat common in CDNs and other hosting providers. Older Python
126 126 versions do not support SNI. Defining this attribute enables clients
127 127 with older Python versions to filter this entry without experiencing
128 128 an opaque SSL failure at connection time.
129 129
130 130 If this is defined, it is important to advertise a non-SNI fallback
131 131 URL or clients running old Python releases may not be able to clone
132 132 with the clonebundles facility.
133 133
134 134 Value should be "true".
135 135
136 136 Manifests can contain multiple entries. Assuming metadata is defined, clients
137 137 will filter entries from the manifest that they don't support. The remaining
138 138 entries are optionally sorted by client preferences
139 139 (``experimental.clonebundleprefers`` config option). The client then attempts
140 140 to fetch the bundle at the first URL in the remaining list.
141 141
142 142 **Errors when downloading a bundle will fail the entire clone operation:
143 143 clients do not automatically fall back to a traditional clone.** The reason
144 144 for this is that if a server is using clone bundles, it is probably doing so
145 145 because the feature is necessary to help it scale. In other words, there
146 146 is an assumption that clone load will be offloaded to another service and
147 147 that the Mercurial server isn't responsible for serving this clone load.
148 148 If that other service experiences issues and clients start mass falling back to
149 149 the original Mercurial server, the added clone load could overwhelm the server
150 150 due to unexpected load and effectively take it offline. Not having clients
151 151 automatically fall back to cloning from the original server mitigates this
152 152 scenario.
153 153
154 154 Because there is no automatic Mercurial server fallback on failure of the
155 155 bundle hosting service, it is important for server operators to view the bundle
156 156 hosting service as an extension of the Mercurial server in terms of
157 157 availability and service level agreements: if the bundle hosting service goes
158 158 down, so does the ability for clients to clone. Note: clients will see a
159 159 message informing them how to bypass the clone bundles facility when a failure
160 160 occurs. So server operators should prepare for some people to follow these
161 161 instructions when a failure occurs, thus driving more load to the original
162 162 Mercurial server when the bundle hosting service fails.
163 163
164 164 The following config options influence the behavior of the clone bundles
165 165 feature:
166 166
167 167 ui.clonebundleadvertise
168 168 Whether the server advertises the existence of the clone bundles feature
169 169 to compatible clients that aren't using it.
170 170
171 171 When this is enabled (the default), a server will send a message to
172 172 compatible clients performing a traditional clone informing them of the
173 173 available clone bundles feature. Compatible clients are those that support
174 174 bundle2 and are advertising support for the clone bundles feature.
175 175
176 176 ui.clonebundlefallback
177 177 Whether to automatically fall back to a traditional clone in case of
178 178 clone bundles failure. Defaults to false for reasons described above.
179 179
180 180 experimental.clonebundles
181 181 Whether the clone bundles feature is enabled on clients. Defaults to true.
182 182
183 183 experimental.clonebundleprefers
184 184 List of "key=value" properties the client prefers in bundles. Downloaded
185 185 bundle manifests will be sorted by the preferences in this list. e.g.
186 186 the value "BUNDLESPEC=gzip-v1, BUNDLESPEC=bzip2=v1" will prefer a gzipped
187 187 version 1 bundle type then bzip2 version 1 bundle type.
188 188
189 189 If not defined, the order in the manifest will be used and the first
190 190 available bundle will be downloaded.
191 191 """
192 192
193 193 from mercurial.i18n import _
194 194 from mercurial.node import nullid
195 195 from mercurial import (
196 196 exchange,
197 197 extensions,
198 198 wireproto,
199 199 )
200 200
201 201 testedwith = 'internal'
202 202
203 203 def capabilities(orig, repo, proto):
204 204 caps = orig(repo, proto)
205 205
206 206 # Only advertise if a manifest exists. This does add some I/O to requests.
207 207 # But this should be cheaper than a wasted network round trip due to
208 208 # missing file.
209 209 if repo.opener.exists('clonebundles.manifest'):
210 210 caps.append('clonebundles')
211 211
212 212 return caps
213 213
214 214 @exchange.getbundle2partsgenerator('clonebundlesadvertise', 0)
215 215 def advertiseclonebundlespart(bundler, repo, source, bundlecaps=None,
216 216 b2caps=None, heads=None, common=None,
217 217 cbattempted=None, **kwargs):
218 218 """Inserts an output part to advertise clone bundles availability."""
219 219 # Allow server operators to disable this behavior.
220 220 # # experimental config: ui.clonebundleadvertise
221 221 if not repo.ui.configbool('ui', 'clonebundleadvertise', True):
222 222 return
223 223
224 224 # Only advertise if a manifest is present.
225 225 if not repo.opener.exists('clonebundles.manifest'):
226 226 return
227 227
228 228 # And when changegroup data is requested.
229 229 if not kwargs.get('cg', True):
230 230 return
231 231
232 232 # And when the client supports clone bundles.
233 233 if cbattempted is None:
234 234 return
235 235
236 236 # And when the client didn't attempt a clone bundle as part of this pull.
237 237 if cbattempted:
238 238 return
239 239
240 240 # And when a full clone is requested.
241 241 # Note: client should not send "cbattempted" for regular pulls. This check
242 242 # is defense in depth.
243 243 if common and common != [nullid]:
244 244 return
245 245
246 246 msg = _('this server supports the experimental "clone bundles" feature '
247 247 'that should enable faster and more reliable cloning\n'
248 248 'help test it by setting the "experimental.clonebundles" config '
249 249 'flag to "true"')
250 250
251 251 bundler.newpart('output', data=msg)
252 252
253 253 def extsetup(ui):
254 254 extensions.wrapfunction(wireproto, '_capabilities', capabilities)
General Comments 0
You need to be logged in to leave comments. Login now