Show More
@@ -1,761 +1,821 | |||
|
1 | 1 | # upgrade.py - functions for in place upgrade of Mercurial repository |
|
2 | 2 | # |
|
3 | 3 | # Copyright (c) 2016-present, Gregory Szorc |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import stat |
|
11 | 11 | import tempfile |
|
12 | 12 | |
|
13 | 13 | from .i18n import _ |
|
14 | 14 | from . import ( |
|
15 | 15 | changelog, |
|
16 | 16 | error, |
|
17 | 17 | localrepo, |
|
18 | 18 | manifest, |
|
19 | 19 | revlog, |
|
20 | 20 | scmutil, |
|
21 | 21 | util, |
|
22 | 22 | vfs as vfsmod, |
|
23 | 23 | ) |
|
24 | 24 | |
|
25 | 25 | def requiredsourcerequirements(repo): |
|
26 | 26 | """Obtain requirements required to be present to upgrade a repo. |
|
27 | 27 | |
|
28 | 28 | An upgrade will not be allowed if the repository doesn't have the |
|
29 | 29 | requirements returned by this function. |
|
30 | 30 | """ |
|
31 | 31 | return set([ |
|
32 | 32 | # Introduced in Mercurial 0.9.2. |
|
33 | 33 | 'revlogv1', |
|
34 | 34 | # Introduced in Mercurial 0.9.2. |
|
35 | 35 | 'store', |
|
36 | 36 | ]) |
|
37 | 37 | |
|
38 | 38 | def blocksourcerequirements(repo): |
|
39 | 39 | """Obtain requirements that will prevent an upgrade from occurring. |
|
40 | 40 | |
|
41 | 41 | An upgrade cannot be performed if the source repository contains a |
|
42 | 42 | requirements in the returned set. |
|
43 | 43 | """ |
|
44 | 44 | return set([ |
|
45 | 45 | # The upgrade code does not yet support these experimental features. |
|
46 | 46 | # This is an artificial limitation. |
|
47 | 47 | 'manifestv2', |
|
48 | 48 | 'treemanifest', |
|
49 | 49 | # This was a precursor to generaldelta and was never enabled by default. |
|
50 | 50 | # It should (hopefully) not exist in the wild. |
|
51 | 51 | 'parentdelta', |
|
52 | 52 | # Upgrade should operate on the actual store, not the shared link. |
|
53 | 53 | 'shared', |
|
54 | 54 | ]) |
|
55 | 55 | |
|
56 | 56 | def supportremovedrequirements(repo): |
|
57 | 57 | """Obtain requirements that can be removed during an upgrade. |
|
58 | 58 | |
|
59 | 59 | If an upgrade were to create a repository that dropped a requirement, |
|
60 | 60 | the dropped requirement must appear in the returned set for the upgrade |
|
61 | 61 | to be allowed. |
|
62 | 62 | """ |
|
63 | 63 | return set() |
|
64 | 64 | |
|
65 | 65 | def supporteddestrequirements(repo): |
|
66 | 66 | """Obtain requirements that upgrade supports in the destination. |
|
67 | 67 | |
|
68 | 68 | If the result of the upgrade would create requirements not in this set, |
|
69 | 69 | the upgrade is disallowed. |
|
70 | 70 | |
|
71 | 71 | Extensions should monkeypatch this to add their custom requirements. |
|
72 | 72 | """ |
|
73 | 73 | return set([ |
|
74 | 74 | 'dotencode', |
|
75 | 75 | 'fncache', |
|
76 | 76 | 'generaldelta', |
|
77 | 77 | 'revlogv1', |
|
78 | 78 | 'store', |
|
79 | 79 | ]) |
|
80 | 80 | |
|
81 | 81 | def allowednewrequirements(repo): |
|
82 | 82 | """Obtain requirements that can be added to a repository during upgrade. |
|
83 | 83 | |
|
84 | 84 | This is used to disallow proposed requirements from being added when |
|
85 | 85 | they weren't present before. |
|
86 | 86 | |
|
87 | 87 | We use a list of allowed requirement additions instead of a list of known |
|
88 | 88 | bad additions because the whitelist approach is safer and will prevent |
|
89 | 89 | future, unknown requirements from accidentally being added. |
|
90 | 90 | """ |
|
91 | 91 | return set([ |
|
92 | 92 | 'dotencode', |
|
93 | 93 | 'fncache', |
|
94 | 94 | 'generaldelta', |
|
95 | 95 | ]) |
|
96 | 96 | |
|
97 | 97 | deficiency = 'deficiency' |
|
98 | 98 | optimisation = 'optimization' |
|
99 | 99 | |
|
100 | 100 | class improvement(object): |
|
101 | 101 | """Represents an improvement that can be made as part of an upgrade. |
|
102 | 102 | |
|
103 | 103 | The following attributes are defined on each instance: |
|
104 | 104 | |
|
105 | 105 | name |
|
106 | 106 | Machine-readable string uniquely identifying this improvement. It |
|
107 | 107 | will be mapped to an action later in the upgrade process. |
|
108 | 108 | |
|
109 | 109 | type |
|
110 | 110 | Either ``deficiency`` or ``optimisation``. A deficiency is an obvious |
|
111 | 111 | problem. An optimization is an action (sometimes optional) that |
|
112 | 112 | can be taken to further improve the state of the repository. |
|
113 | 113 | |
|
114 | 114 | description |
|
115 | 115 | Message intended for humans explaining the improvement in more detail, |
|
116 | 116 | including the implications of it. For ``deficiency`` types, should be |
|
117 | 117 | worded in the present tense. For ``optimisation`` types, should be |
|
118 | 118 | worded in the future tense. |
|
119 | 119 | |
|
120 | 120 | upgrademessage |
|
121 | 121 | Message intended for humans explaining what an upgrade addressing this |
|
122 | 122 | issue will do. Should be worded in the future tense. |
|
123 | 123 | """ |
|
124 | 124 | def __init__(self, name, type, description, upgrademessage): |
|
125 | 125 | self.name = name |
|
126 | 126 | self.type = type |
|
127 | 127 | self.description = description |
|
128 | 128 | self.upgrademessage = upgrademessage |
|
129 | 129 | |
|
130 | 130 | def __eq__(self, other): |
|
131 | 131 | if not isinstance(other, improvement): |
|
132 | 132 | # This is what python tell use to do |
|
133 | 133 | return NotImplemented |
|
134 | 134 | return self.name == other.name |
|
135 | 135 | |
|
136 | 136 | def __ne__(self, other): |
|
137 | 137 | return not self == other |
|
138 | 138 | |
|
139 | 139 | def __hash__(self): |
|
140 | 140 | return hash(self.name) |
|
141 | 141 | |
|
142 | 142 | class formatvariant(improvement): |
|
143 | """an improvement subclass dedicated to repository format | |
|
143 | """an improvement subclass dedicated to repository format""" | |
|
144 | type = deficiency | |
|
145 | ### The following attributes should be defined for each class: | |
|
146 | ||
|
147 | # machine-readable string uniquely identifying this improvement. it will be | |
|
148 | # mapped to an action later in the upgrade process. | |
|
149 | name = None | |
|
144 | 150 | |
|
145 | extra attributes: | |
|
151 | # message intended for humans explaining the improvement in more detail, | |
|
152 | # including the implications of it ``deficiency`` types, should be worded | |
|
153 | # in the present tense. | |
|
154 | description = None | |
|
155 | ||
|
156 | # message intended for humans explaining what an upgrade addressing this | |
|
157 | # issue will do. should be worded in the future tense. | |
|
158 | upgrademessage = None | |
|
146 | 159 | |
|
147 | fromdefault (``deficiency`` types only) | |
|
148 | Boolean indicating whether the current (deficient) state deviates | |
|
149 | from Mercurial's default configuration. | |
|
160 | # value of current Mercurial default for new repository | |
|
161 | default = None | |
|
162 | ||
|
163 | def __init__(self): | |
|
164 | raise NotImplementedError() | |
|
165 | ||
|
166 | @staticmethod | |
|
167 | def fromrepo(repo): | |
|
168 | """current value of the variant in the repository""" | |
|
169 | raise NotImplementedError() | |
|
150 | 170 | |
|
151 | fromconfig (``deficiency`` types only) | |
|
152 | Boolean indicating whether the current (deficient) state deviates | |
|
153 |
|
|
|
171 | @staticmethod | |
|
172 | def fromconfig(repo): | |
|
173 | """current value of the variant in the configuration""" | |
|
174 | raise NotImplementedError() | |
|
175 | ||
|
176 | class requirementformatvariant(formatvariant): | |
|
177 | """formatvariant based on a 'requirement' name. | |
|
178 | ||
|
179 | Many format variant are controlled by a 'requirement'. We define a small | |
|
180 | subclass to factor the code. | |
|
154 | 181 | """ |
|
155 | 182 | |
|
156 | def __init__(self, name, description, upgrademessage, fromdefault, | |
|
157 | fromconfig): | |
|
158 | super(formatvariant, self).__init__(name, deficiency, description, | |
|
159 | upgrademessage) | |
|
160 | self.fromdefault = fromdefault | |
|
161 | self.fromconfig = fromconfig | |
|
183 | # the requirement that control this format variant | |
|
184 | _requirement = None | |
|
185 | ||
|
186 | @staticmethod | |
|
187 | def _newreporequirements(repo): | |
|
188 | return localrepo.newreporequirements(repo) | |
|
189 | ||
|
190 | @classmethod | |
|
191 | def fromrepo(cls, repo): | |
|
192 | assert cls._requirement is not None | |
|
193 | return cls._requirement in repo.requirements | |
|
194 | ||
|
195 | @classmethod | |
|
196 | def fromconfig(cls, repo): | |
|
197 | assert cls._requirement is not None | |
|
198 | return cls._requirement in cls._newreporequirements(repo) | |
|
199 | ||
|
200 | class fncache(requirementformatvariant): | |
|
201 | name = 'fncache' | |
|
202 | ||
|
203 | _requirement = 'fncache' | |
|
204 | ||
|
205 | default = True | |
|
206 | ||
|
207 | description = _('long and reserved filenames may not work correctly; ' | |
|
208 | 'repository performance is sub-optimal') | |
|
209 | ||
|
210 | upgrademessage = _('repository will be more resilient to storing ' | |
|
211 | 'certain paths and performance of certain ' | |
|
212 | 'operations should be improved') | |
|
213 | ||
|
214 | class dotencode(requirementformatvariant): | |
|
215 | name = 'dotencode' | |
|
216 | ||
|
217 | _requirement = 'dotencode' | |
|
218 | ||
|
219 | default = True | |
|
220 | ||
|
221 | description = _('storage of filenames beginning with a period or ' | |
|
222 | 'space may not work correctly') | |
|
223 | ||
|
224 | upgrademessage = _('repository will be better able to store files ' | |
|
225 | 'beginning with a space or period') | |
|
226 | ||
|
227 | class generaldelta(requirementformatvariant): | |
|
228 | name = 'generaldelta' | |
|
229 | ||
|
230 | _requirement = 'generaldelta' | |
|
231 | ||
|
232 | default = True | |
|
233 | ||
|
234 | description = _('deltas within internal storage are unable to ' | |
|
235 | 'choose optimal revisions; repository is larger and ' | |
|
236 | 'slower than it could be; interaction with other ' | |
|
237 | 'repositories may require extra network and CPU ' | |
|
238 | 'resources, making "hg push" and "hg pull" slower') | |
|
239 | ||
|
240 | upgrademessage = _('repository storage will be able to create ' | |
|
241 | 'optimal deltas; new repository data will be ' | |
|
242 | 'smaller and read times should decrease; ' | |
|
243 | 'interacting with other repositories using this ' | |
|
244 | 'storage model should require less network and ' | |
|
245 | 'CPU resources, making "hg push" and "hg pull" ' | |
|
246 | 'faster') | |
|
247 | ||
|
248 | class removecldeltachain(formatvariant): | |
|
249 | name = 'removecldeltachain' | |
|
250 | ||
|
251 | default = True | |
|
252 | ||
|
253 | description = _('changelog storage is using deltas instead of ' | |
|
254 | 'raw entries; changelog reading and any ' | |
|
255 | 'operation relying on changelog data are slower ' | |
|
256 | 'than they could be') | |
|
257 | ||
|
258 | upgrademessage = _('changelog storage will be reformated to ' | |
|
259 | 'store raw entries; changelog reading will be ' | |
|
260 | 'faster; changelog size may be reduced') | |
|
261 | ||
|
262 | @staticmethod | |
|
263 | def fromrepo(repo): | |
|
264 | # Mercurial 4.0 changed changelogs to not use delta chains. Search for | |
|
265 | # changelogs with deltas. | |
|
266 | cl = repo.changelog | |
|
267 | chainbase = cl.chainbase | |
|
268 | return all(rev == chainbase(rev) for rev in cl) | |
|
269 | ||
|
270 | @staticmethod | |
|
271 | def fromconfig(repo): | |
|
272 | return True | |
|
162 | 273 | |
|
163 | 274 | def finddeficiencies(repo): |
|
164 | 275 | """returns a list of deficiencies that the repo suffer from""" |
|
165 | newreporeqs = localrepo.newreporequirements(repo) | |
|
166 | ||
|
167 | 276 | deficiencies = [] |
|
168 | 277 | |
|
169 | 278 | # We could detect lack of revlogv1 and store here, but they were added |
|
170 | 279 | # in 0.9.2 and we don't support upgrading repos without these |
|
171 | 280 | # requirements, so let's not bother. |
|
172 | 281 | |
|
173 | if 'fncache' not in repo.requirements: | |
|
174 |
deficiencies.append(f |
|
|
175 | name='fncache', | |
|
176 | description=_('long and reserved filenames may not work correctly; ' | |
|
177 | 'repository performance is sub-optimal'), | |
|
178 | upgrademessage=_('repository will be more resilient to storing ' | |
|
179 | 'certain paths and performance of certain ' | |
|
180 | 'operations should be improved'), | |
|
181 | fromdefault=True, | |
|
182 | fromconfig='fncache' in newreporeqs)) | |
|
183 | ||
|
184 | if 'dotencode' not in repo.requirements: | |
|
185 | deficiencies.append(formatvariant( | |
|
186 | name='dotencode', | |
|
187 | description=_('storage of filenames beginning with a period or ' | |
|
188 | 'space may not work correctly'), | |
|
189 | upgrademessage=_('repository will be better able to store files ' | |
|
190 | 'beginning with a space or period'), | |
|
191 | fromdefault=True, | |
|
192 | fromconfig='dotencode' in newreporeqs)) | |
|
193 | ||
|
194 | if 'generaldelta' not in repo.requirements: | |
|
195 | deficiencies.append(formatvariant( | |
|
196 | name='generaldelta', | |
|
197 | description=_('deltas within internal storage are unable to ' | |
|
198 | 'choose optimal revisions; repository is larger and ' | |
|
199 | 'slower than it could be; interaction with other ' | |
|
200 | 'repositories may require extra network and CPU ' | |
|
201 | 'resources, making "hg push" and "hg pull" slower'), | |
|
202 | upgrademessage=_('repository storage will be able to create ' | |
|
203 | 'optimal deltas; new repository data will be ' | |
|
204 | 'smaller and read times should decrease; ' | |
|
205 | 'interacting with other repositories using this ' | |
|
206 | 'storage model should require less network and ' | |
|
207 | 'CPU resources, making "hg push" and "hg pull" ' | |
|
208 | 'faster'), | |
|
209 | fromdefault=True, | |
|
210 | fromconfig='generaldelta' in newreporeqs)) | |
|
211 | ||
|
212 | # Mercurial 4.0 changed changelogs to not use delta chains. Search for | |
|
213 | # changelogs with deltas. | |
|
214 | cl = repo.changelog | |
|
215 | for rev in cl: | |
|
216 | chainbase = cl.chainbase(rev) | |
|
217 | if chainbase != rev: | |
|
218 | deficiencies.append(formatvariant( | |
|
219 | name='removecldeltachain', | |
|
220 | description=_('changelog storage is using deltas instead of ' | |
|
221 | 'raw entries; changelog reading and any ' | |
|
222 | 'operation relying on changelog data are slower ' | |
|
223 | 'than they could be'), | |
|
224 | upgrademessage=_('changelog storage will be reformated to ' | |
|
225 | 'store raw entries; changelog reading will be ' | |
|
226 | 'faster; changelog size may be reduced'), | |
|
227 | fromdefault=True, | |
|
228 | fromconfig=True)) | |
|
229 | break | |
|
282 | if not fncache.fromrepo(repo): | |
|
283 | deficiencies.append(fncache) | |
|
284 | if not dotencode.fromrepo(repo): | |
|
285 | deficiencies.append(dotencode) | |
|
286 | if not generaldelta.fromrepo(repo): | |
|
287 | deficiencies.append(generaldelta) | |
|
288 | if not removecldeltachain.fromrepo(repo): | |
|
289 | deficiencies.append(removecldeltachain) | |
|
230 | 290 | |
|
231 | 291 | return deficiencies |
|
232 | 292 | |
|
233 | 293 | def findoptimizations(repo): |
|
234 | 294 | """Determine optimisation that could be used during upgrade""" |
|
235 | 295 | # These are unconditionally added. There is logic later that figures out |
|
236 | 296 | # which ones to apply. |
|
237 | 297 | optimizations = [] |
|
238 | 298 | |
|
239 | 299 | optimizations.append(improvement( |
|
240 | 300 | name='redeltaparent', |
|
241 | 301 | type=optimisation, |
|
242 | 302 | description=_('deltas within internal storage will be recalculated to ' |
|
243 | 303 | 'choose an optimal base revision where this was not ' |
|
244 | 304 | 'already done; the size of the repository may shrink and ' |
|
245 | 305 | 'various operations may become faster; the first time ' |
|
246 | 306 | 'this optimization is performed could slow down upgrade ' |
|
247 | 307 | 'execution considerably; subsequent invocations should ' |
|
248 | 308 | 'not run noticeably slower'), |
|
249 | 309 | upgrademessage=_('deltas within internal storage will choose a new ' |
|
250 | 310 | 'base revision if needed'))) |
|
251 | 311 | |
|
252 | 312 | optimizations.append(improvement( |
|
253 | 313 | name='redeltamultibase', |
|
254 | 314 | type=optimisation, |
|
255 | 315 | description=_('deltas within internal storage will be recalculated ' |
|
256 | 316 | 'against multiple base revision and the smallest ' |
|
257 | 317 | 'difference will be used; the size of the repository may ' |
|
258 | 318 | 'shrink significantly when there are many merges; this ' |
|
259 | 319 | 'optimization will slow down execution in proportion to ' |
|
260 | 320 | 'the number of merges in the repository and the amount ' |
|
261 | 321 | 'of files in the repository; this slow down should not ' |
|
262 | 322 | 'be significant unless there are tens of thousands of ' |
|
263 | 323 | 'files and thousands of merges'), |
|
264 | 324 | upgrademessage=_('deltas within internal storage will choose an ' |
|
265 | 325 | 'optimal delta by computing deltas against multiple ' |
|
266 | 326 | 'parents; may slow down execution time ' |
|
267 | 327 | 'significantly'))) |
|
268 | 328 | |
|
269 | 329 | optimizations.append(improvement( |
|
270 | 330 | name='redeltaall', |
|
271 | 331 | type=optimisation, |
|
272 | 332 | description=_('deltas within internal storage will always be ' |
|
273 | 333 | 'recalculated without reusing prior deltas; this will ' |
|
274 | 334 | 'likely make execution run several times slower; this ' |
|
275 | 335 | 'optimization is typically not needed'), |
|
276 | 336 | upgrademessage=_('deltas within internal storage will be fully ' |
|
277 | 337 | 'recomputed; this will likely drastically slow down ' |
|
278 | 338 | 'execution time'))) |
|
279 | 339 | |
|
280 | 340 | return optimizations |
|
281 | 341 | |
|
282 | 342 | def determineactions(repo, deficiencies, sourcereqs, destreqs): |
|
283 | 343 | """Determine upgrade actions that will be performed. |
|
284 | 344 | |
|
285 | 345 | Given a list of improvements as returned by ``finddeficiencies`` and |
|
286 | 346 | ``findoptimizations``, determine the list of upgrade actions that |
|
287 | 347 | will be performed. |
|
288 | 348 | |
|
289 | 349 | The role of this function is to filter improvements if needed, apply |
|
290 | 350 | recommended optimizations from the improvements list that make sense, |
|
291 | 351 | etc. |
|
292 | 352 | |
|
293 | 353 | Returns a list of action names. |
|
294 | 354 | """ |
|
295 | 355 | newactions = [] |
|
296 | 356 | |
|
297 | 357 | knownreqs = supporteddestrequirements(repo) |
|
298 | 358 | |
|
299 | 359 | for d in deficiencies: |
|
300 | 360 | name = d.name |
|
301 | 361 | |
|
302 | 362 | # If the action is a requirement that doesn't show up in the |
|
303 | 363 | # destination requirements, prune the action. |
|
304 | 364 | if name in knownreqs and name not in destreqs: |
|
305 | 365 | continue |
|
306 | 366 | |
|
307 | 367 | newactions.append(d) |
|
308 | 368 | |
|
309 | 369 | # FUTURE consider adding some optimizations here for certain transitions. |
|
310 | 370 | # e.g. adding generaldelta could schedule parent redeltas. |
|
311 | 371 | |
|
312 | 372 | return newactions |
|
313 | 373 | |
|
314 | 374 | def _revlogfrompath(repo, path): |
|
315 | 375 | """Obtain a revlog from a repo path. |
|
316 | 376 | |
|
317 | 377 | An instance of the appropriate class is returned. |
|
318 | 378 | """ |
|
319 | 379 | if path == '00changelog.i': |
|
320 | 380 | return changelog.changelog(repo.svfs) |
|
321 | 381 | elif path.endswith('00manifest.i'): |
|
322 | 382 | mandir = path[:-len('00manifest.i')] |
|
323 | 383 | return manifest.manifestrevlog(repo.svfs, dir=mandir) |
|
324 | 384 | else: |
|
325 | 385 | # Filelogs don't do anything special with settings. So we can use a |
|
326 | 386 | # vanilla revlog. |
|
327 | 387 | return revlog.revlog(repo.svfs, path) |
|
328 | 388 | |
|
329 | 389 | def _copyrevlogs(ui, srcrepo, dstrepo, tr, deltareuse, aggressivemergedeltas): |
|
330 | 390 | """Copy revlogs between 2 repos.""" |
|
331 | 391 | revcount = 0 |
|
332 | 392 | srcsize = 0 |
|
333 | 393 | srcrawsize = 0 |
|
334 | 394 | dstsize = 0 |
|
335 | 395 | fcount = 0 |
|
336 | 396 | frevcount = 0 |
|
337 | 397 | fsrcsize = 0 |
|
338 | 398 | frawsize = 0 |
|
339 | 399 | fdstsize = 0 |
|
340 | 400 | mcount = 0 |
|
341 | 401 | mrevcount = 0 |
|
342 | 402 | msrcsize = 0 |
|
343 | 403 | mrawsize = 0 |
|
344 | 404 | mdstsize = 0 |
|
345 | 405 | crevcount = 0 |
|
346 | 406 | csrcsize = 0 |
|
347 | 407 | crawsize = 0 |
|
348 | 408 | cdstsize = 0 |
|
349 | 409 | |
|
350 | 410 | # Perform a pass to collect metadata. This validates we can open all |
|
351 | 411 | # source files and allows a unified progress bar to be displayed. |
|
352 | 412 | for unencoded, encoded, size in srcrepo.store.walk(): |
|
353 | 413 | if unencoded.endswith('.d'): |
|
354 | 414 | continue |
|
355 | 415 | |
|
356 | 416 | rl = _revlogfrompath(srcrepo, unencoded) |
|
357 | 417 | revcount += len(rl) |
|
358 | 418 | |
|
359 | 419 | datasize = 0 |
|
360 | 420 | rawsize = 0 |
|
361 | 421 | idx = rl.index |
|
362 | 422 | for rev in rl: |
|
363 | 423 | e = idx[rev] |
|
364 | 424 | datasize += e[1] |
|
365 | 425 | rawsize += e[2] |
|
366 | 426 | |
|
367 | 427 | srcsize += datasize |
|
368 | 428 | srcrawsize += rawsize |
|
369 | 429 | |
|
370 | 430 | # This is for the separate progress bars. |
|
371 | 431 | if isinstance(rl, changelog.changelog): |
|
372 | 432 | crevcount += len(rl) |
|
373 | 433 | csrcsize += datasize |
|
374 | 434 | crawsize += rawsize |
|
375 | 435 | elif isinstance(rl, manifest.manifestrevlog): |
|
376 | 436 | mcount += 1 |
|
377 | 437 | mrevcount += len(rl) |
|
378 | 438 | msrcsize += datasize |
|
379 | 439 | mrawsize += rawsize |
|
380 | 440 | elif isinstance(rl, revlog.revlog): |
|
381 | 441 | fcount += 1 |
|
382 | 442 | frevcount += len(rl) |
|
383 | 443 | fsrcsize += datasize |
|
384 | 444 | frawsize += rawsize |
|
385 | 445 | |
|
386 | 446 | if not revcount: |
|
387 | 447 | return |
|
388 | 448 | |
|
389 | 449 | ui.write(_('migrating %d total revisions (%d in filelogs, %d in manifests, ' |
|
390 | 450 | '%d in changelog)\n') % |
|
391 | 451 | (revcount, frevcount, mrevcount, crevcount)) |
|
392 | 452 | ui.write(_('migrating %s in store; %s tracked data\n') % ( |
|
393 | 453 | (util.bytecount(srcsize), util.bytecount(srcrawsize)))) |
|
394 | 454 | |
|
395 | 455 | # Used to keep track of progress. |
|
396 | 456 | progress = [] |
|
397 | 457 | def oncopiedrevision(rl, rev, node): |
|
398 | 458 | progress[1] += 1 |
|
399 | 459 | srcrepo.ui.progress(progress[0], progress[1], total=progress[2]) |
|
400 | 460 | |
|
401 | 461 | # Do the actual copying. |
|
402 | 462 | # FUTURE this operation can be farmed off to worker processes. |
|
403 | 463 | seen = set() |
|
404 | 464 | for unencoded, encoded, size in srcrepo.store.walk(): |
|
405 | 465 | if unencoded.endswith('.d'): |
|
406 | 466 | continue |
|
407 | 467 | |
|
408 | 468 | oldrl = _revlogfrompath(srcrepo, unencoded) |
|
409 | 469 | newrl = _revlogfrompath(dstrepo, unencoded) |
|
410 | 470 | |
|
411 | 471 | if isinstance(oldrl, changelog.changelog) and 'c' not in seen: |
|
412 | 472 | ui.write(_('finished migrating %d manifest revisions across %d ' |
|
413 | 473 | 'manifests; change in size: %s\n') % |
|
414 | 474 | (mrevcount, mcount, util.bytecount(mdstsize - msrcsize))) |
|
415 | 475 | |
|
416 | 476 | ui.write(_('migrating changelog containing %d revisions ' |
|
417 | 477 | '(%s in store; %s tracked data)\n') % |
|
418 | 478 | (crevcount, util.bytecount(csrcsize), |
|
419 | 479 | util.bytecount(crawsize))) |
|
420 | 480 | seen.add('c') |
|
421 | 481 | progress[:] = [_('changelog revisions'), 0, crevcount] |
|
422 | 482 | elif isinstance(oldrl, manifest.manifestrevlog) and 'm' not in seen: |
|
423 | 483 | ui.write(_('finished migrating %d filelog revisions across %d ' |
|
424 | 484 | 'filelogs; change in size: %s\n') % |
|
425 | 485 | (frevcount, fcount, util.bytecount(fdstsize - fsrcsize))) |
|
426 | 486 | |
|
427 | 487 | ui.write(_('migrating %d manifests containing %d revisions ' |
|
428 | 488 | '(%s in store; %s tracked data)\n') % |
|
429 | 489 | (mcount, mrevcount, util.bytecount(msrcsize), |
|
430 | 490 | util.bytecount(mrawsize))) |
|
431 | 491 | seen.add('m') |
|
432 | 492 | progress[:] = [_('manifest revisions'), 0, mrevcount] |
|
433 | 493 | elif 'f' not in seen: |
|
434 | 494 | ui.write(_('migrating %d filelogs containing %d revisions ' |
|
435 | 495 | '(%s in store; %s tracked data)\n') % |
|
436 | 496 | (fcount, frevcount, util.bytecount(fsrcsize), |
|
437 | 497 | util.bytecount(frawsize))) |
|
438 | 498 | seen.add('f') |
|
439 | 499 | progress[:] = [_('file revisions'), 0, frevcount] |
|
440 | 500 | |
|
441 | 501 | ui.progress(progress[0], progress[1], total=progress[2]) |
|
442 | 502 | |
|
443 | 503 | ui.note(_('cloning %d revisions from %s\n') % (len(oldrl), unencoded)) |
|
444 | 504 | oldrl.clone(tr, newrl, addrevisioncb=oncopiedrevision, |
|
445 | 505 | deltareuse=deltareuse, |
|
446 | 506 | aggressivemergedeltas=aggressivemergedeltas) |
|
447 | 507 | |
|
448 | 508 | datasize = 0 |
|
449 | 509 | idx = newrl.index |
|
450 | 510 | for rev in newrl: |
|
451 | 511 | datasize += idx[rev][1] |
|
452 | 512 | |
|
453 | 513 | dstsize += datasize |
|
454 | 514 | |
|
455 | 515 | if isinstance(newrl, changelog.changelog): |
|
456 | 516 | cdstsize += datasize |
|
457 | 517 | elif isinstance(newrl, manifest.manifestrevlog): |
|
458 | 518 | mdstsize += datasize |
|
459 | 519 | else: |
|
460 | 520 | fdstsize += datasize |
|
461 | 521 | |
|
462 | 522 | ui.progress(progress[0], None) |
|
463 | 523 | |
|
464 | 524 | ui.write(_('finished migrating %d changelog revisions; change in size: ' |
|
465 | 525 | '%s\n') % (crevcount, util.bytecount(cdstsize - csrcsize))) |
|
466 | 526 | |
|
467 | 527 | ui.write(_('finished migrating %d total revisions; total change in store ' |
|
468 | 528 | 'size: %s\n') % (revcount, util.bytecount(dstsize - srcsize))) |
|
469 | 529 | |
|
470 | 530 | def _filterstorefile(srcrepo, dstrepo, requirements, path, mode, st): |
|
471 | 531 | """Determine whether to copy a store file during upgrade. |
|
472 | 532 | |
|
473 | 533 | This function is called when migrating store files from ``srcrepo`` to |
|
474 | 534 | ``dstrepo`` as part of upgrading a repository. |
|
475 | 535 | |
|
476 | 536 | Args: |
|
477 | 537 | srcrepo: repo we are copying from |
|
478 | 538 | dstrepo: repo we are copying to |
|
479 | 539 | requirements: set of requirements for ``dstrepo`` |
|
480 | 540 | path: store file being examined |
|
481 | 541 | mode: the ``ST_MODE`` file type of ``path`` |
|
482 | 542 | st: ``stat`` data structure for ``path`` |
|
483 | 543 | |
|
484 | 544 | Function should return ``True`` if the file is to be copied. |
|
485 | 545 | """ |
|
486 | 546 | # Skip revlogs. |
|
487 | 547 | if path.endswith(('.i', '.d')): |
|
488 | 548 | return False |
|
489 | 549 | # Skip transaction related files. |
|
490 | 550 | if path.startswith('undo'): |
|
491 | 551 | return False |
|
492 | 552 | # Only copy regular files. |
|
493 | 553 | if mode != stat.S_IFREG: |
|
494 | 554 | return False |
|
495 | 555 | # Skip other skipped files. |
|
496 | 556 | if path in ('lock', 'fncache'): |
|
497 | 557 | return False |
|
498 | 558 | |
|
499 | 559 | return True |
|
500 | 560 | |
|
501 | 561 | def _finishdatamigration(ui, srcrepo, dstrepo, requirements): |
|
502 | 562 | """Hook point for extensions to perform additional actions during upgrade. |
|
503 | 563 | |
|
504 | 564 | This function is called after revlogs and store files have been copied but |
|
505 | 565 | before the new store is swapped into the original location. |
|
506 | 566 | """ |
|
507 | 567 | |
|
508 | 568 | def _upgraderepo(ui, srcrepo, dstrepo, requirements, actions): |
|
509 | 569 | """Do the low-level work of upgrading a repository. |
|
510 | 570 | |
|
511 | 571 | The upgrade is effectively performed as a copy between a source |
|
512 | 572 | repository and a temporary destination repository. |
|
513 | 573 | |
|
514 | 574 | The source repository is unmodified for as long as possible so the |
|
515 | 575 | upgrade can abort at any time without causing loss of service for |
|
516 | 576 | readers and without corrupting the source repository. |
|
517 | 577 | """ |
|
518 | 578 | assert srcrepo.currentwlock() |
|
519 | 579 | assert dstrepo.currentwlock() |
|
520 | 580 | |
|
521 | 581 | ui.write(_('(it is safe to interrupt this process any time before ' |
|
522 | 582 | 'data migration completes)\n')) |
|
523 | 583 | |
|
524 | 584 | if 'redeltaall' in actions: |
|
525 | 585 | deltareuse = revlog.revlog.DELTAREUSENEVER |
|
526 | 586 | elif 'redeltaparent' in actions: |
|
527 | 587 | deltareuse = revlog.revlog.DELTAREUSESAMEREVS |
|
528 | 588 | elif 'redeltamultibase' in actions: |
|
529 | 589 | deltareuse = revlog.revlog.DELTAREUSESAMEREVS |
|
530 | 590 | else: |
|
531 | 591 | deltareuse = revlog.revlog.DELTAREUSEALWAYS |
|
532 | 592 | |
|
533 | 593 | with dstrepo.transaction('upgrade') as tr: |
|
534 | 594 | _copyrevlogs(ui, srcrepo, dstrepo, tr, deltareuse, |
|
535 | 595 | 'redeltamultibase' in actions) |
|
536 | 596 | |
|
537 | 597 | # Now copy other files in the store directory. |
|
538 | 598 | # The sorted() makes execution deterministic. |
|
539 | 599 | for p, kind, st in sorted(srcrepo.store.vfs.readdir('', stat=True)): |
|
540 | 600 | if not _filterstorefile(srcrepo, dstrepo, requirements, |
|
541 | 601 | p, kind, st): |
|
542 | 602 | continue |
|
543 | 603 | |
|
544 | 604 | srcrepo.ui.write(_('copying %s\n') % p) |
|
545 | 605 | src = srcrepo.store.rawvfs.join(p) |
|
546 | 606 | dst = dstrepo.store.rawvfs.join(p) |
|
547 | 607 | util.copyfile(src, dst, copystat=True) |
|
548 | 608 | |
|
549 | 609 | _finishdatamigration(ui, srcrepo, dstrepo, requirements) |
|
550 | 610 | |
|
551 | 611 | ui.write(_('data fully migrated to temporary repository\n')) |
|
552 | 612 | |
|
553 | 613 | backuppath = tempfile.mkdtemp(prefix='upgradebackup.', dir=srcrepo.path) |
|
554 | 614 | backupvfs = vfsmod.vfs(backuppath) |
|
555 | 615 | |
|
556 | 616 | # Make a backup of requires file first, as it is the first to be modified. |
|
557 | 617 | util.copyfile(srcrepo.vfs.join('requires'), backupvfs.join('requires')) |
|
558 | 618 | |
|
559 | 619 | # We install an arbitrary requirement that clients must not support |
|
560 | 620 | # as a mechanism to lock out new clients during the data swap. This is |
|
561 | 621 | # better than allowing a client to continue while the repository is in |
|
562 | 622 | # an inconsistent state. |
|
563 | 623 | ui.write(_('marking source repository as being upgraded; clients will be ' |
|
564 | 624 | 'unable to read from repository\n')) |
|
565 | 625 | scmutil.writerequires(srcrepo.vfs, |
|
566 | 626 | srcrepo.requirements | set(['upgradeinprogress'])) |
|
567 | 627 | |
|
568 | 628 | ui.write(_('starting in-place swap of repository data\n')) |
|
569 | 629 | ui.write(_('replaced files will be backed up at %s\n') % |
|
570 | 630 | backuppath) |
|
571 | 631 | |
|
572 | 632 | # Now swap in the new store directory. Doing it as a rename should make |
|
573 | 633 | # the operation nearly instantaneous and atomic (at least in well-behaved |
|
574 | 634 | # environments). |
|
575 | 635 | ui.write(_('replacing store...\n')) |
|
576 | 636 | tstart = util.timer() |
|
577 | 637 | util.rename(srcrepo.spath, backupvfs.join('store')) |
|
578 | 638 | util.rename(dstrepo.spath, srcrepo.spath) |
|
579 | 639 | elapsed = util.timer() - tstart |
|
580 | 640 | ui.write(_('store replacement complete; repository was inconsistent for ' |
|
581 | 641 | '%0.1fs\n') % elapsed) |
|
582 | 642 | |
|
583 | 643 | # We first write the requirements file. Any new requirements will lock |
|
584 | 644 | # out legacy clients. |
|
585 | 645 | ui.write(_('finalizing requirements file and making repository readable ' |
|
586 | 646 | 'again\n')) |
|
587 | 647 | scmutil.writerequires(srcrepo.vfs, requirements) |
|
588 | 648 | |
|
589 | 649 | # The lock file from the old store won't be removed because nothing has a |
|
590 | 650 | # reference to its new location. So clean it up manually. Alternatively, we |
|
591 | 651 | # could update srcrepo.svfs and other variables to point to the new |
|
592 | 652 | # location. This is simpler. |
|
593 | 653 | backupvfs.unlink('store/lock') |
|
594 | 654 | |
|
595 | 655 | return backuppath |
|
596 | 656 | |
|
597 | 657 | def upgraderepo(ui, repo, run=False, optimize=None): |
|
598 | 658 | """Upgrade a repository in place.""" |
|
599 | 659 | optimize = set(optimize or []) |
|
600 | 660 | repo = repo.unfiltered() |
|
601 | 661 | |
|
602 | 662 | # Ensure the repository can be upgraded. |
|
603 | 663 | missingreqs = requiredsourcerequirements(repo) - repo.requirements |
|
604 | 664 | if missingreqs: |
|
605 | 665 | raise error.Abort(_('cannot upgrade repository; requirement ' |
|
606 | 666 | 'missing: %s') % _(', ').join(sorted(missingreqs))) |
|
607 | 667 | |
|
608 | 668 | blockedreqs = blocksourcerequirements(repo) & repo.requirements |
|
609 | 669 | if blockedreqs: |
|
610 | 670 | raise error.Abort(_('cannot upgrade repository; unsupported source ' |
|
611 | 671 | 'requirement: %s') % |
|
612 | 672 | _(', ').join(sorted(blockedreqs))) |
|
613 | 673 | |
|
614 | 674 | # FUTURE there is potentially a need to control the wanted requirements via |
|
615 | 675 | # command arguments or via an extension hook point. |
|
616 | 676 | newreqs = localrepo.newreporequirements(repo) |
|
617 | 677 | |
|
618 | 678 | noremovereqs = (repo.requirements - newreqs - |
|
619 | 679 | supportremovedrequirements(repo)) |
|
620 | 680 | if noremovereqs: |
|
621 | 681 | raise error.Abort(_('cannot upgrade repository; requirement would be ' |
|
622 | 682 | 'removed: %s') % _(', ').join(sorted(noremovereqs))) |
|
623 | 683 | |
|
624 | 684 | noaddreqs = (newreqs - repo.requirements - |
|
625 | 685 | allowednewrequirements(repo)) |
|
626 | 686 | if noaddreqs: |
|
627 | 687 | raise error.Abort(_('cannot upgrade repository; do not support adding ' |
|
628 | 688 | 'requirement: %s') % |
|
629 | 689 | _(', ').join(sorted(noaddreqs))) |
|
630 | 690 | |
|
631 | 691 | unsupportedreqs = newreqs - supporteddestrequirements(repo) |
|
632 | 692 | if unsupportedreqs: |
|
633 | 693 | raise error.Abort(_('cannot upgrade repository; do not support ' |
|
634 | 694 | 'destination requirement: %s') % |
|
635 | 695 | _(', ').join(sorted(unsupportedreqs))) |
|
636 | 696 | |
|
637 | 697 | # Find and validate all improvements that can be made. |
|
638 | 698 | alloptimizations = findoptimizations(repo) |
|
639 | 699 | |
|
640 | 700 | # Apply and Validate arguments. |
|
641 | 701 | optimizations = [] |
|
642 | 702 | for o in alloptimizations: |
|
643 | 703 | if o.name in optimize: |
|
644 | 704 | optimizations.append(o) |
|
645 | 705 | optimize.discard(o.name) |
|
646 | 706 | |
|
647 | 707 | if optimize: # anything left is unknown |
|
648 | 708 | raise error.Abort(_('unknown optimization action requested: %s') % |
|
649 | 709 | ', '.join(sorted(optimize)), |
|
650 | 710 | hint=_('run without arguments to see valid ' |
|
651 | 711 | 'optimizations')) |
|
652 | 712 | |
|
653 | 713 | deficiencies = finddeficiencies(repo) |
|
654 | 714 | actions = determineactions(repo, deficiencies, repo.requirements, newreqs) |
|
655 | 715 | actions.extend(o for o in sorted(optimizations) |
|
656 | 716 | # determineactions could have added optimisation |
|
657 | 717 | if o not in actions) |
|
658 | 718 | |
|
659 | 719 | def printrequirements(): |
|
660 | 720 | ui.write(_('requirements\n')) |
|
661 | 721 | ui.write(_(' preserved: %s\n') % |
|
662 | 722 | _(', ').join(sorted(newreqs & repo.requirements))) |
|
663 | 723 | |
|
664 | 724 | if repo.requirements - newreqs: |
|
665 | 725 | ui.write(_(' removed: %s\n') % |
|
666 | 726 | _(', ').join(sorted(repo.requirements - newreqs))) |
|
667 | 727 | |
|
668 | 728 | if newreqs - repo.requirements: |
|
669 | 729 | ui.write(_(' added: %s\n') % |
|
670 | 730 | _(', ').join(sorted(newreqs - repo.requirements))) |
|
671 | 731 | |
|
672 | 732 | ui.write('\n') |
|
673 | 733 | |
|
674 | 734 | def printupgradeactions(): |
|
675 | 735 | for a in actions: |
|
676 | 736 | ui.write('%s\n %s\n\n' % (a.name, a.upgrademessage)) |
|
677 | 737 | |
|
678 | 738 | if not run: |
|
679 | 739 | fromconfig = [] |
|
680 | 740 | onlydefault = [] |
|
681 | 741 | |
|
682 | 742 | for d in deficiencies: |
|
683 | if d.fromconfig: | |
|
743 | if d.fromconfig(repo): | |
|
684 | 744 | fromconfig.append(d) |
|
685 |
elif d. |
|
|
745 | elif d.default: | |
|
686 | 746 | onlydefault.append(d) |
|
687 | 747 | |
|
688 | 748 | if fromconfig or onlydefault: |
|
689 | 749 | |
|
690 | 750 | if fromconfig: |
|
691 | 751 | ui.write(_('repository lacks features recommended by ' |
|
692 | 752 | 'current config options:\n\n')) |
|
693 | 753 | for i in fromconfig: |
|
694 | 754 | ui.write('%s\n %s\n\n' % (i.name, i.description)) |
|
695 | 755 | |
|
696 | 756 | if onlydefault: |
|
697 | 757 | ui.write(_('repository lacks features used by the default ' |
|
698 | 758 | 'config options:\n\n')) |
|
699 | 759 | for i in onlydefault: |
|
700 | 760 | ui.write('%s\n %s\n\n' % (i.name, i.description)) |
|
701 | 761 | |
|
702 | 762 | ui.write('\n') |
|
703 | 763 | else: |
|
704 | 764 | ui.write(_('(no feature deficiencies found in existing ' |
|
705 | 765 | 'repository)\n')) |
|
706 | 766 | |
|
707 | 767 | ui.write(_('performing an upgrade with "--run" will make the following ' |
|
708 | 768 | 'changes:\n\n')) |
|
709 | 769 | |
|
710 | 770 | printrequirements() |
|
711 | 771 | printupgradeactions() |
|
712 | 772 | |
|
713 | 773 | unusedoptimize = [i for i in alloptimizations if i not in actions] |
|
714 | 774 | |
|
715 | 775 | if unusedoptimize: |
|
716 | 776 | ui.write(_('additional optimizations are available by specifying ' |
|
717 | 777 | '"--optimize <name>":\n\n')) |
|
718 | 778 | for i in unusedoptimize: |
|
719 | 779 | ui.write(_('%s\n %s\n\n') % (i.name, i.description)) |
|
720 | 780 | return |
|
721 | 781 | |
|
722 | 782 | # Else we're in the run=true case. |
|
723 | 783 | ui.write(_('upgrade will perform the following actions:\n\n')) |
|
724 | 784 | printrequirements() |
|
725 | 785 | printupgradeactions() |
|
726 | 786 | |
|
727 | 787 | upgradeactions = [a.name for a in actions] |
|
728 | 788 | |
|
729 | 789 | ui.write(_('beginning upgrade...\n')) |
|
730 | 790 | with repo.wlock(): |
|
731 | 791 | with repo.lock(): |
|
732 | 792 | ui.write(_('repository locked and read-only\n')) |
|
733 | 793 | # Our strategy for upgrading the repository is to create a new, |
|
734 | 794 | # temporary repository, write data to it, then do a swap of the |
|
735 | 795 | # data. There are less heavyweight ways to do this, but it is easier |
|
736 | 796 | # to create a new repo object than to instantiate all the components |
|
737 | 797 | # (like the store) separately. |
|
738 | 798 | tmppath = tempfile.mkdtemp(prefix='upgrade.', dir=repo.path) |
|
739 | 799 | backuppath = None |
|
740 | 800 | try: |
|
741 | 801 | ui.write(_('creating temporary repository to stage migrated ' |
|
742 | 802 | 'data: %s\n') % tmppath) |
|
743 | 803 | dstrepo = localrepo.localrepository(repo.baseui, |
|
744 | 804 | path=tmppath, |
|
745 | 805 | create=True) |
|
746 | 806 | |
|
747 | 807 | with dstrepo.wlock(): |
|
748 | 808 | with dstrepo.lock(): |
|
749 | 809 | backuppath = _upgraderepo(ui, repo, dstrepo, newreqs, |
|
750 | 810 | upgradeactions) |
|
751 | 811 | |
|
752 | 812 | finally: |
|
753 | 813 | ui.write(_('removing temporary repository %s\n') % tmppath) |
|
754 | 814 | repo.vfs.rmtree(tmppath, forcibly=True) |
|
755 | 815 | |
|
756 | 816 | if backuppath: |
|
757 | 817 | ui.warn(_('copy of old repository backed up at %s\n') % |
|
758 | 818 | backuppath) |
|
759 | 819 | ui.warn(_('the old repository will not be deleted; remove ' |
|
760 | 820 | 'it to free up disk space once the upgraded ' |
|
761 | 821 | 'repository is verified\n')) |
General Comments 0
You need to be logged in to leave comments.
Login now