[Notes] [Git][BuildStream/buildstream][raoul/870-root-cache-dir] CASQuota: Move cache check methods into new Class



Title: GitLab

Raoul Hidalgo Charman pushed to branch raoul/870-root-cache-dir at BuildStream / buildstream

Commits:

11 changed files:

Changes:

  • buildstream/_artifactcache.py
    ... ... @@ -22,7 +22,7 @@ import os
    22 22
     from collections.abc import Mapping
    
    23 23
     
    
    24 24
     from .types import _KeyStrength
    
    25
    -from ._exceptions import ArtifactError, CASError, LoadError, LoadErrorReason
    
    25
    +from ._exceptions import ArtifactError, CASError
    
    26 26
     from ._message import Message, MessageType
    
    27 27
     from . import utils
    
    28 28
     from . import _yaml
    
    ... ... @@ -46,39 +46,6 @@ class ArtifactCacheSpec(CASRemoteSpec):
    46 46
         pass
    
    47 47
     
    
    48 48
     
    
    49
    -# ArtifactCacheUsage
    
    50
    -#
    
    51
    -# A simple object to report the current artifact cache
    
    52
    -# usage details.
    
    53
    -#
    
    54
    -# Note that this uses the user configured cache quota
    
    55
    -# rather than the internal quota with protective headroom
    
    56
    -# removed, to provide a more sensible value to display to
    
    57
    -# the user.
    
    58
    -#
    
    59
    -# Args:
    
    60
    -#    artifacts (ArtifactCache): The artifact cache to get the status of
    
    61
    -#
    
    62
    -class ArtifactCacheUsage():
    
    63
    -
    
    64
    -    def __init__(self, artifacts):
    
    65
    -        context = artifacts.context
    
    66
    -        self.quota_config = context.config_cache_quota       # Configured quota
    
    67
    -        self.quota_size = artifacts._cache_quota_original    # Resolved cache quota in bytes
    
    68
    -        self.used_size = artifacts.get_cache_size()          # Size used by artifacts in bytes
    
    69
    -        self.used_percent = 0                                # Percentage of the quota used
    
    70
    -        if self.quota_size is not None:
    
    71
    -            self.used_percent = int(self.used_size * 100 / self.quota_size)
    
    72
    -
    
    73
    -    # Formattable into a human readable string
    
    74
    -    #
    
    75
    -    def __str__(self):
    
    76
    -        return "{} / {} ({}%)" \
    
    77
    -            .format(utils._pretty_size(self.used_size, dec_places=1),
    
    78
    -                    self.quota_config,
    
    79
    -                    self.used_percent)
    
    80
    -
    
    81
    -
    
    82 49
     # An ArtifactCache manages artifacts.
    
    83 50
     #
    
    84 51
     # Args:
    
    ... ... @@ -90,16 +57,14 @@ class ArtifactCache():
    90 57
             self.extractdir = context.extractdir
    
    91 58
     
    
    92 59
             self.cas = context.get_cascache()
    
    60
    +        self.casquota = context.get_casquota()
    
    61
    +        self.casquota._calculate_cache_quota()
    
    93 62
     
    
    94 63
             self.global_remote_specs = []
    
    95 64
             self.project_remote_specs = {}
    
    96 65
     
    
    97 66
             self._required_elements = set()       # The elements required for this session
    
    98
    -        self._cache_size = None               # The current cache size, sometimes it's an estimate
    
    99
    -        self._cache_quota = None              # The cache quota
    
    100
    -        self._cache_quota_original = None     # The cache quota as specified by the user, in bytes
    
    101
    -        self._cache_quota_headroom = None     # The headroom in bytes before reaching the quota or full disk
    
    102
    -        self._cache_lower_threshold = None    # The target cache size for a cleanup
    
    67
    +
    
    103 68
             self._remotes_setup = False           # Check to prevent double-setup of remotes
    
    104 69
     
    
    105 70
             # Per-project list of _CASRemote instances.
    
    ... ... @@ -110,8 +75,6 @@ class ArtifactCache():
    110 75
     
    
    111 76
             os.makedirs(self.extractdir, exist_ok=True)
    
    112 77
     
    
    113
    -        self._calculate_cache_quota()
    
    114
    -
    
    115 78
         # setup_remotes():
    
    116 79
         #
    
    117 80
         # Sets up which remotes to use
    
    ... ... @@ -235,7 +198,7 @@ class ArtifactCache():
    235 198
             space_saved = 0
    
    236 199
     
    
    237 200
             # Start off with an announcement with as much info as possible
    
    238
    -        volume_size, volume_avail = self._get_cache_volume_size()
    
    201
    +        volume_size, volume_avail = self.casquota._get_cache_volume_size()
    
    239 202
             self._message(MessageType.STATUS, "Starting cache cleanup",
    
    240 203
                           detail=("Elements required by the current build plan: {}\n" +
    
    241 204
                                   "User specified quota: {} ({})\n" +
    
    ... ... @@ -243,8 +206,8 @@ class ArtifactCache():
    243 206
                                   "Cache volume: {} total, {} available")
    
    244 207
                           .format(len(self._required_elements),
    
    245 208
                                   context.config_cache_quota,
    
    246
    -                              utils._pretty_size(self._cache_quota_original, dec_places=2),
    
    247
    -                              utils._pretty_size(self.get_cache_size(), dec_places=2),
    
    209
    +                              utils._pretty_size(self.casquota._cache_quota, dec_places=2),
    
    210
    +                              utils._pretty_size(self.casquota.get_cache_size(), dec_places=2),
    
    248 211
                                   utils._pretty_size(volume_size, dec_places=2),
    
    249 212
                                   utils._pretty_size(volume_avail, dec_places=2)))
    
    250 213
     
    
    ... ... @@ -261,9 +224,11 @@ class ArtifactCache():
    261 224
                 ])
    
    262 225
     
    
    263 226
             # Do a real computation of the cache size once, just in case
    
    264
    -        self.compute_cache_size()
    
    227
    +        self.casquota.compute_cache_size()
    
    228
    +        usage = CASCacheUsage(self.casquota)
    
    229
    +        self._message(MessageType.STATUS, "Cache usage recomputed: {}".format(usage))
    
    265 230
     
    
    266
    -        while self.get_cache_size() >= self._cache_lower_threshold:
    
    231
    +        while self.casquota.get_cache_size() >= self.casquota._cache_lower_threshold:
    
    267 232
                 try:
    
    268 233
                     to_remove = artifacts.pop(0)
    
    269 234
                 except IndexError:
    
    ... ... @@ -280,7 +245,7 @@ class ArtifactCache():
    280 245
                               "Please increase the cache-quota in {} and/or make more disk space."
    
    281 246
                               .format(removed_ref_count,
    
    282 247
                                       utils._pretty_size(space_saved, dec_places=2),
    
    283
    -                                  utils._pretty_size(self.get_cache_size(), dec_places=2),
    
    248
    +                                  utils._pretty_size(self.casquota.get_cache_size(), dec_places=2),
    
    284 249
                                       len(self._required_elements),
    
    285 250
                                       (context.config_origin or default_conf)))
    
    286 251
     
    
    ... ... @@ -306,7 +271,7 @@ class ArtifactCache():
    306 271
                                       to_remove))
    
    307 272
     
    
    308 273
                     # Remove the size from the removed size
    
    309
    -                self.set_cache_size(self._cache_size - size)
    
    274
    +                self.casquota.set_cache_size(self.casquota._cache_size - size)
    
    310 275
     
    
    311 276
                     # User callback
    
    312 277
                     #
    
    ... ... @@ -322,29 +287,12 @@ class ArtifactCache():
    322 287
                                   "Cache usage is now: {}")
    
    323 288
                           .format(removed_ref_count,
    
    324 289
                                   utils._pretty_size(space_saved, dec_places=2),
    
    325
    -                              utils._pretty_size(self.get_cache_size(), dec_places=2)))
    
    326
    -
    
    327
    -        return self.get_cache_size()
    
    328
    -
    
    329
    -    # compute_cache_size()
    
    330
    -    #
    
    331
    -    # Computes the real artifact cache size by calling
    
    332
    -    # the abstract calculate_cache_size() method.
    
    333
    -    #
    
    334
    -    # Returns:
    
    335
    -    #    (int): The size of the artifact cache.
    
    336
    -    #
    
    337
    -    def compute_cache_size(self):
    
    338
    -        old_cache_size = self._cache_size
    
    339
    -        new_cache_size = self.cas.calculate_cache_size()
    
    340
    -
    
    341
    -        if old_cache_size != new_cache_size:
    
    342
    -            self._cache_size = new_cache_size
    
    290
    +                              utils._pretty_size(self.casquota.get_cache_size(), dec_places=2)))
    
    343 291
     
    
    344
    -            usage = ArtifactCacheUsage(self)
    
    345
    -            self._message(MessageType.STATUS, "Cache usage recomputed: {}".format(usage))
    
    292
    +        return self.casquota.get_cache_size()
    
    346 293
     
    
    347
    -        return self._cache_size
    
    294
    +    def full(self):
    
    295
    +        return self.casquota.full()
    
    348 296
     
    
    349 297
         # add_artifact_size()
    
    350 298
         #
    
    ... ... @@ -355,71 +303,10 @@ class ArtifactCache():
    355 303
         #     artifact_size (int): The size to add.
    
    356 304
         #
    
    357 305
         def add_artifact_size(self, artifact_size):
    
    358
    -        cache_size = self.get_cache_size()
    
    306
    +        cache_size = self.casquota.get_cache_size()
    
    359 307
             cache_size += artifact_size
    
    360 308
     
    
    361
    -        self.set_cache_size(cache_size)
    
    362
    -
    
    363
    -    # get_cache_size()
    
    364
    -    #
    
    365
    -    # Fetches the cached size of the cache, this is sometimes
    
    366
    -    # an estimate and periodically adjusted to the real size
    
    367
    -    # when a cache size calculation job runs.
    
    368
    -    #
    
    369
    -    # When it is an estimate, the value is either correct, or
    
    370
    -    # it is greater than the actual cache size.
    
    371
    -    #
    
    372
    -    # Returns:
    
    373
    -    #     (int) An approximation of the artifact cache size, in bytes.
    
    374
    -    #
    
    375
    -    def get_cache_size(self):
    
    376
    -
    
    377
    -        # If we don't currently have an estimate, figure out the real cache size.
    
    378
    -        if self._cache_size is None:
    
    379
    -            stored_size = self._read_cache_size()
    
    380
    -            if stored_size is not None:
    
    381
    -                self._cache_size = stored_size
    
    382
    -            else:
    
    383
    -                self.compute_cache_size()
    
    384
    -
    
    385
    -        return self._cache_size
    
    386
    -
    
    387
    -    # set_cache_size()
    
    388
    -    #
    
    389
    -    # Forcefully set the overall cache size.
    
    390
    -    #
    
    391
    -    # This is used to update the size in the main process after
    
    392
    -    # having calculated in a cleanup or a cache size calculation job.
    
    393
    -    #
    
    394
    -    # Args:
    
    395
    -    #     cache_size (int): The size to set.
    
    396
    -    #
    
    397
    -    def set_cache_size(self, cache_size):
    
    398
    -
    
    399
    -        assert cache_size is not None
    
    400
    -
    
    401
    -        self._cache_size = cache_size
    
    402
    -        self._write_cache_size(self._cache_size)
    
    403
    -
    
    404
    -    # full()
    
    405
    -    #
    
    406
    -    # Checks if the artifact cache is full, either
    
    407
    -    # because the user configured quota has been exceeded
    
    408
    -    # or because the underlying disk is almost full.
    
    409
    -    #
    
    410
    -    # Returns:
    
    411
    -    #    (bool): True if the artifact cache is full
    
    412
    -    #
    
    413
    -    def full(self):
    
    414
    -
    
    415
    -        if self.get_cache_size() > self._cache_quota:
    
    416
    -            return True
    
    417
    -
    
    418
    -        _, volume_avail = self._get_cache_volume_size()
    
    419
    -        if volume_avail < self._cache_quota_headroom:
    
    420
    -            return True
    
    421
    -
    
    422
    -        return False
    
    309
    +        self.casquota.set_cache_size(cache_size)
    
    423 310
     
    
    424 311
         # preflight():
    
    425 312
         #
    
    ... ... @@ -882,142 +769,6 @@ class ArtifactCache():
    882 769
             with self.context.timed_activity("Initializing remote caches", silent_nested=True):
    
    883 770
                 self.initialize_remotes(on_failure=remote_failed)
    
    884 771
     
    
    885
    -    # _write_cache_size()
    
    886
    -    #
    
    887
    -    # Writes the given size of the artifact to the cache's size file
    
    888
    -    #
    
    889
    -    # Args:
    
    890
    -    #    size (int): The size of the artifact cache to record
    
    891
    -    #
    
    892
    -    def _write_cache_size(self, size):
    
    893
    -        assert isinstance(size, int)
    
    894
    -        size_file_path = os.path.join(self.context.casdir, CACHE_SIZE_FILE)
    
    895
    -        with utils.save_file_atomic(size_file_path, "w") as f:
    
    896
    -            f.write(str(size))
    
    897
    -
    
    898
    -    # _read_cache_size()
    
    899
    -    #
    
    900
    -    # Reads and returns the size of the artifact cache that's stored in the
    
    901
    -    # cache's size file
    
    902
    -    #
    
    903
    -    # Returns:
    
    904
    -    #    (int): The size of the artifact cache, as recorded in the file
    
    905
    -    #
    
    906
    -    def _read_cache_size(self):
    
    907
    -        size_file_path = os.path.join(self.context.casdir, CACHE_SIZE_FILE)
    
    908
    -
    
    909
    -        if not os.path.exists(size_file_path):
    
    910
    -            return None
    
    911
    -
    
    912
    -        with open(size_file_path, "r") as f:
    
    913
    -            size = f.read()
    
    914
    -
    
    915
    -        try:
    
    916
    -            num_size = int(size)
    
    917
    -        except ValueError as e:
    
    918
    -            raise ArtifactError("Size '{}' parsed from '{}' was not an integer".format(
    
    919
    -                size, size_file_path)) from e
    
    920
    -
    
    921
    -        return num_size
    
    922
    -
    
    923
    -    # _calculate_cache_quota()
    
    924
    -    #
    
    925
    -    # Calculates and sets the cache quota and lower threshold based on the
    
    926
    -    # quota set in Context.
    
    927
    -    # It checks that the quota is both a valid _expression_, and that there is
    
    928
    -    # enough disk space to satisfy that quota
    
    929
    -    #
    
    930
    -    def _calculate_cache_quota(self):
    
    931
    -        # Headroom intended to give BuildStream a bit of leeway.
    
    932
    -        # This acts as the minimum size of cache_quota and also
    
    933
    -        # is taken from the user requested cache_quota.
    
    934
    -        #
    
    935
    -        if 'BST_TEST_SUITE' in os.environ:
    
    936
    -            self._cache_quota_headroom = 0
    
    937
    -        else:
    
    938
    -            self._cache_quota_headroom = 2e9
    
    939
    -
    
    940
    -        try:
    
    941
    -            cache_quota = utils._parse_size(self.context.config_cache_quota,
    
    942
    -                                            self.context.casdir)
    
    943
    -        except utils.UtilError as e:
    
    944
    -            raise LoadError(LoadErrorReason.INVALID_DATA,
    
    945
    -                            "{}\nPlease specify the value in bytes or as a % of full disk space.\n"
    
    946
    -                            "\nValid values are, for example: 800M 10G 1T 50%\n"
    
    947
    -                            .format(str(e))) from e
    
    948
    -
    
    949
    -        total_size, available_space = self._get_cache_volume_size()
    
    950
    -        cache_size = self.get_cache_size()
    
    951
    -
    
    952
    -        # Ensure system has enough storage for the cache_quota
    
    953
    -        #
    
    954
    -        # If cache_quota is none, set it to the maximum it could possibly be.
    
    955
    -        #
    
    956
    -        # Also check that cache_quota is at least as large as our headroom.
    
    957
    -        #
    
    958
    -        if cache_quota is None:  # Infinity, set to max system storage
    
    959
    -            cache_quota = cache_size + available_space
    
    960
    -        if cache_quota < self._cache_quota_headroom:  # Check minimum
    
    961
    -            raise LoadError(LoadErrorReason.INVALID_DATA,
    
    962
    -                            "Invalid cache quota ({}): ".format(utils._pretty_size(cache_quota)) +
    
    963
    -                            "BuildStream requires a minimum cache quota of 2G.")
    
    964
    -        elif cache_quota > total_size:
    
    965
    -            # A quota greater than the total disk size is certianly an error
    
    966
    -            raise ArtifactError("Your system does not have enough available " +
    
    967
    -                                "space to support the cache quota specified.",
    
    968
    -                                detail=("You have specified a quota of {quota} total disk space.\n" +
    
    969
    -                                        "The filesystem containing {local_cache_path} only " +
    
    970
    -                                        "has {total_size} total disk space.")
    
    971
    -                                .format(
    
    972
    -                                    quota=self.context.config_cache_quota,
    
    973
    -                                    local_cache_path=self.context.casdir,
    
    974
    -                                    total_size=utils._pretty_size(total_size)),
    
    975
    -                                reason='insufficient-storage-for-quota')
    
    976
    -        elif cache_quota > cache_size + available_space:
    
    977
    -            # The quota does not fit in the available space, this is a warning
    
    978
    -            if '%' in self.context.config_cache_quota:
    
    979
    -                available = (available_space / total_size) * 100
    
    980
    -                available = '{}% of total disk space'.format(round(available, 1))
    
    981
    -            else:
    
    982
    -                available = utils._pretty_size(available_space)
    
    983
    -
    
    984
    -            self._message(MessageType.WARN,
    
    985
    -                          "Your system does not have enough available " +
    
    986
    -                          "space to support the cache quota specified.",
    
    987
    -                          detail=("You have specified a quota of {quota} total disk space.\n" +
    
    988
    -                                  "The filesystem containing {local_cache_path} only " +
    
    989
    -                                  "has {available_size} available.")
    
    990
    -                          .format(quota=self.context.config_cache_quota,
    
    991
    -                                  local_cache_path=self.context.casdir,
    
    992
    -                                  available_size=available))
    
    993
    -
    
    994
    -        # Place a slight headroom (2e9 (2GB) on the cache_quota) into
    
    995
    -        # cache_quota to try and avoid exceptions.
    
    996
    -        #
    
    997
    -        # Of course, we might still end up running out during a build
    
    998
    -        # if we end up writing more than 2G, but hey, this stuff is
    
    999
    -        # already really fuzzy.
    
    1000
    -        #
    
    1001
    -        self._cache_quota_original = cache_quota
    
    1002
    -        self._cache_quota = cache_quota - self._cache_quota_headroom
    
    1003
    -        self._cache_lower_threshold = self._cache_quota / 2
    
    1004
    -
    
    1005
    -    # _get_cache_volume_size()
    
    1006
    -    #
    
    1007
    -    # Get the available space and total space for the volume on
    
    1008
    -    # which the artifact cache is located.
    
    1009
    -    #
    
    1010
    -    # Returns:
    
    1011
    -    #    (int): The total number of bytes on the volume
    
    1012
    -    #    (int): The number of available bytes on the volume
    
    1013
    -    #
    
    1014
    -    # NOTE: We use this stub to allow the test cases
    
    1015
    -    #       to override what an artifact cache thinks
    
    1016
    -    #       about it's disk size and available bytes.
    
    1017
    -    #
    
    1018
    -    def _get_cache_volume_size(self):
    
    1019
    -        return utils._get_volume_size(self.context.casdir)
    
    1020
    -
    
    1021 772
     
    
    1022 773
     # _configured_remote_artifact_cache_specs():
    
    1023 774
     #
    

  • buildstream/_cas/__init__.py
    ... ... @@ -17,5 +17,5 @@
    17 17
     #  Authors:
    
    18 18
     #        Tristan Van Berkom <tristan vanberkom codethink co uk>
    
    19 19
     
    
    20
    -from .cascache import CASCache
    
    20
    +from .cascache import CASCache, CASQuota, CASCacheUsage
    
    21 21
     from .casremote import CASRemote, CASRemoteSpec

  • buildstream/_cas/cascache.py
    ... ... @@ -32,17 +32,53 @@ from .._protos.build.bazel.remote.execution.v2 import remote_execution_pb2
    32 32
     from .._protos.buildstream.v2 import buildstream_pb2
    
    33 33
     
    
    34 34
     from .. import utils
    
    35
    -from .._exceptions import CASCacheError
    
    35
    +from .._exceptions import CASCacheError, LoadError, LoadErrorReason
    
    36
    +from .._message import Message, MessageType
    
    36 37
     
    
    37 38
     from .casremote import BlobNotFound, _CASBatchRead, _CASBatchUpdate
    
    38 39
     
    
    39 40
     _BUFFER_SIZE = 65536
    
    40 41
     
    
    41 42
     
    
    43
    +CACHE_SIZE_FILE = "cache_size"
    
    44
    +
    
    45
    +
    
    46
    +# CASCacheUsage
    
    47
    +#
    
    48
    +# A simple object to report the current CAS cache usage details.
    
    49
    +#
    
    50
    +# Note that this uses the user configured cache quota
    
    51
    +# rather than the internal quota with protective headroom
    
    52
    +# removed, to provide a more sensible value to display to
    
    53
    +# the user.
    
    54
    +#
    
    55
    +# Args:
    
    56
    +#    cas (CASQuota): The CAS cache to get the status of
    
    57
    +#
    
    58
    +class CASCacheUsage():
    
    59
    +
    
    60
    +    def __init__(self, casquota):
    
    61
    +        self.quota_config = casquota._config_cache_quota          # Configured quota
    
    62
    +        self.quota_size = casquota._cache_quota_original          # Resolved cache quota in bytes
    
    63
    +        self.used_size = casquota.get_cache_size()                # Size used by artifacts in bytes
    
    64
    +        self.used_percent = 0                                # Percentage of the quota used
    
    65
    +        if self.quota_size is not None:
    
    66
    +            self.used_percent = int(self.used_size * 100 / self.quota_size)
    
    67
    +
    
    68
    +    # Formattable into a human readable string
    
    69
    +    #
    
    70
    +    def __str__(self):
    
    71
    +        return "{} / {} ({}%)" \
    
    72
    +            .format(utils._pretty_size(self.used_size, dec_places=1),
    
    73
    +                    self.quota_config,
    
    74
    +                    self.used_percent)
    
    75
    +
    
    76
    +
    
    42 77
     # A CASCache manages a CAS repository as specified in the Remote Execution API.
    
    43 78
     #
    
    44 79
     # Args:
    
    45 80
     #     path (str): The root directory for the CAS repository
    
    81
    +#     cache_quota (int): User configured cache quota
    
    46 82
     #
    
    47 83
     class CASCache():
    
    48 84
     
    
    ... ... @@ -459,16 +495,6 @@ class CASCache():
    459 495
             except FileNotFoundError as e:
    
    460 496
                 raise CASCacheError("Attempt to access unavailable ref: {}".format(e)) from e
    
    461 497
     
    
    462
    -    # calculate_cache_size()
    
    463
    -    #
    
    464
    -    # Return the real disk usage of the CAS cache.
    
    465
    -    #
    
    466
    -    # Returns:
    
    467
    -    #    (int): The size of the cache.
    
    468
    -    #
    
    469
    -    def calculate_cache_size(self):
    
    470
    -        return utils._get_dir_size(self.casdir)
    
    471
    -
    
    472 498
         # list_refs():
    
    473 499
         #
    
    474 500
         # List refs in Least Recently Modified (LRM) order.
    
    ... ... @@ -1043,6 +1069,248 @@ class CASCache():
    1043 1069
             batch.send()
    
    1044 1070
     
    
    1045 1071
     
    
    1072
    +class CASQuota:
    
    1073
    +    def __init__(self, context):
    
    1074
    +        self.cas = context.get_cascache()
    
    1075
    +        self.casdir = self.cas.casdir
    
    1076
    +        self._config_cache_quota = context.config_cache_quota
    
    1077
    +        self._config_cache_quota_string = context.config_cache_quota_string
    
    1078
    +        self._cache_size = None               # The current cache size, sometimes it's an estimate
    
    1079
    +        self._cache_quota = None              # The cache quota
    
    1080
    +        self._cache_quota_original = None     # The cache quota as specified by the user, in bytes
    
    1081
    +        self._cache_quota_headroom = None     # The headroom in bytes before reaching the quota or full disk
    
    1082
    +        self._cache_lower_threshold = None    # The target cache size for a cleanup
    
    1083
    +        self.available_space = None
    
    1084
    +
    
    1085
    +        self._message = context.message
    
    1086
    +
    
    1087
    +        self._calculate_cache_quota()
    
    1088
    +
    
    1089
    +    # compute_cache_size()
    
    1090
    +    #
    
    1091
    +    # Computes the real artifact cache size by calling
    
    1092
    +    # the abstract calculate_cache_size() method.
    
    1093
    +    #
    
    1094
    +    # Returns:
    
    1095
    +    #    (int): The size of the artifact cache.
    
    1096
    +    #
    
    1097
    +    def compute_cache_size(self):
    
    1098
    +        old_cache_size = self._cache_size
    
    1099
    +        new_cache_size = self.calculate_cache_size()
    
    1100
    +
    
    1101
    +        if old_cache_size != new_cache_size:
    
    1102
    +            self._cache_size = new_cache_size
    
    1103
    +
    
    1104
    +        return self._cache_size
    
    1105
    +
    
    1106
    +    # calculate_cache_size()
    
    1107
    +    #
    
    1108
    +    # Return the real disk usage of the CAS cache.
    
    1109
    +    #
    
    1110
    +    # Returns:
    
    1111
    +    #    (int): The size of the cache.
    
    1112
    +    #
    
    1113
    +    def calculate_cache_size(self):
    
    1114
    +        return utils._get_dir_size(self.casdir)
    
    1115
    +
    
    1116
    +    # get_cache_size()
    
    1117
    +    #
    
    1118
    +    # Fetches the cached size of the cache, this is sometimes
    
    1119
    +    # an estimate and periodically adjusted to the real size
    
    1120
    +    # when a cache size calculation job runs.
    
    1121
    +    #
    
    1122
    +    # When it is an estimate, the value is either correct, or
    
    1123
    +    # it is greater than the actual cache size.
    
    1124
    +    #
    
    1125
    +    # Returns:
    
    1126
    +    #     (int) An approximation of the artifact cache size, in bytes.
    
    1127
    +    #
    
    1128
    +    def get_cache_size(self):
    
    1129
    +
    
    1130
    +        # If we don't currently have an estimate, figure out the real cache size.
    
    1131
    +        if self._cache_size is None:
    
    1132
    +            stored_size = self._read_cache_size()
    
    1133
    +            if stored_size is not None:
    
    1134
    +                self._cache_size = stored_size
    
    1135
    +            else:
    
    1136
    +                self._cache_size = self.compute_cache_size()
    
    1137
    +
    
    1138
    +        return self._cache_size
    
    1139
    +
    
    1140
    +    # set_cache_size()
    
    1141
    +    #
    
    1142
    +    # Forcefully set the overall cache size.
    
    1143
    +    #
    
    1144
    +    # This is used to update the size in the main process after
    
    1145
    +    # having calculated in a cleanup or a cache size calculation job.
    
    1146
    +    #
    
    1147
    +    # Args:
    
    1148
    +    #     cache_size (int): The size to set.
    
    1149
    +    #
    
    1150
    +    def set_cache_size(self, cache_size):
    
    1151
    +
    
    1152
    +        assert cache_size is not None
    
    1153
    +
    
    1154
    +        self._cache_size = cache_size
    
    1155
    +        self._write_cache_size(self._cache_size)
    
    1156
    +
    
    1157
    +    # full()
    
    1158
    +    #
    
    1159
    +    # Checks if the artifact cache is full, either
    
    1160
    +    # because the user configured quota has been exceeded
    
    1161
    +    # or because the underlying disk is almost full.
    
    1162
    +    #
    
    1163
    +    # Returns:
    
    1164
    +    #    (bool): True if the artifact cache is full
    
    1165
    +    #
    
    1166
    +    def full(self):
    
    1167
    +
    
    1168
    +        if self.get_cache_size() > self._cache_quota:
    
    1169
    +            return True
    
    1170
    +
    
    1171
    +        _, volume_avail = self._get_cache_volume_size()
    
    1172
    +        if volume_avail < self._cache_quota_headroom:
    
    1173
    +            return True
    
    1174
    +
    
    1175
    +        return False
    
    1176
    +
    
    1177
    +    ################################################
    
    1178
    +    #             Local Private Methods            #
    
    1179
    +    ################################################
    
    1180
    +
    
    1181
    +    # _read_cache_size()
    
    1182
    +    #
    
    1183
    +    # Reads and returns the size of the artifact cache that's stored in the
    
    1184
    +    # cache's size file
    
    1185
    +    #
    
    1186
    +    # Returns:
    
    1187
    +    #    (int): The size of the artifact cache, as recorded in the file
    
    1188
    +    #
    
    1189
    +    def _read_cache_size(self):
    
    1190
    +        size_file_path = os.path.join(self.casdir, CACHE_SIZE_FILE)
    
    1191
    +
    
    1192
    +        if not os.path.exists(size_file_path):
    
    1193
    +            return None
    
    1194
    +
    
    1195
    +        with open(size_file_path, "r") as f:
    
    1196
    +            size = f.read()
    
    1197
    +
    
    1198
    +        try:
    
    1199
    +            num_size = int(size)
    
    1200
    +        except ValueError as e:
    
    1201
    +            raise CASCacheError("Size '{}' parsed from '{}' was not an integer".format(
    
    1202
    +                size, size_file_path)) from e
    
    1203
    +
    
    1204
    +        return num_size
    
    1205
    +
    
    1206
    +    # _write_cache_size()
    
    1207
    +    #
    
    1208
    +    # Writes the given size of the artifact to the cache's size file
    
    1209
    +    #
    
    1210
    +    # Args:
    
    1211
    +    #    size (int): The size of the artifact cache to record
    
    1212
    +    #
    
    1213
    +    def _write_cache_size(self, size):
    
    1214
    +        assert isinstance(size, int)
    
    1215
    +        size_file_path = os.path.join(self.casdir, CACHE_SIZE_FILE)
    
    1216
    +        with utils.save_file_atomic(size_file_path, "w") as f:
    
    1217
    +            f.write(str(size))
    
    1218
    +
    
    1219
    +    # _get_cache_volume_size()
    
    1220
    +    #
    
    1221
    +    # Get the available space and total space for the volume on
    
    1222
    +    # which the artifact cache is located.
    
    1223
    +    #
    
    1224
    +    # Returns:
    
    1225
    +    #    (int): The total number of bytes on the volume
    
    1226
    +    #    (int): The number of available bytes on the volume
    
    1227
    +    #
    
    1228
    +    # NOTE: We use this stub to allow the test cases
    
    1229
    +    #       to override what an artifact cache thinks
    
    1230
    +    #       about it's disk size and available bytes.
    
    1231
    +    #
    
    1232
    +    def _get_cache_volume_size(self):
    
    1233
    +        return utils._get_volume_size(self.casdir)
    
    1234
    +
    
    1235
    +    # _calculate_cache_quota()
    
    1236
    +    #
    
    1237
    +    # Calculates and sets the cache quota and lower threshold based on the
    
    1238
    +    # quota set in Context.
    
    1239
    +    # It checks that the quota is both a valid _expression_, and that there is
    
    1240
    +    # enough disk space to satisfy that quota
    
    1241
    +    #
    
    1242
    +    def _calculate_cache_quota(self):
    
    1243
    +        # Headroom intended to give BuildStream a bit of leeway.
    
    1244
    +        # This acts as the minimum size of cache_quota and also
    
    1245
    +        # is taken from the user requested cache_quota.
    
    1246
    +        #
    
    1247
    +        if 'BST_TEST_SUITE' in os.environ:
    
    1248
    +            self._cache_quota_headroom = 0
    
    1249
    +        else:
    
    1250
    +            self._cache_quota_headroom = 2e9
    
    1251
    +
    
    1252
    +        total_size, available_space = self._get_cache_volume_size()
    
    1253
    +        cache_size = self.get_cache_size()
    
    1254
    +        self.available_space = available_space
    
    1255
    +
    
    1256
    +        # Ensure system has enough storage for the cache_quota
    
    1257
    +        #
    
    1258
    +        # If cache_quota is none, set it to the maximum it could possibly be.
    
    1259
    +        #
    
    1260
    +        # Also check that cache_quota is at least as large as our headroom.
    
    1261
    +        #
    
    1262
    +        cache_quota = self._config_cache_quota
    
    1263
    +        if cache_quota is None:  # Infinity, set to max system storage
    
    1264
    +            cache_quota = cache_size + available_space
    
    1265
    +        if cache_quota < self._cache_quota_headroom:  # Check minimum
    
    1266
    +            raise LoadError(LoadErrorReason.INVALID_DATA,
    
    1267
    +                            "Invalid cache quota ({}): ".format(utils._pretty_size(cache_quota)) +
    
    1268
    +                            "BuildStream requires a minimum cache quota of 2G.")
    
    1269
    +        elif cache_quota > total_size:
    
    1270
    +            # A quota greater than the total disk size is certianly an error
    
    1271
    +            raise CASCacheError("Your system does not have enough available " +
    
    1272
    +                                "space to support the cache quota specified.",
    
    1273
    +                                detail=("You have specified a quota of {quota} total disk space.\n" +
    
    1274
    +                                        "The filesystem containing {local_cache_path} only " +
    
    1275
    +                                        "has {total_size} total disk space.")
    
    1276
    +                                .format(
    
    1277
    +                                    quota=self._config_cache_quota,
    
    1278
    +                                    local_cache_path=self.casdir,
    
    1279
    +                                    total_size=utils._pretty_size(total_size)),
    
    1280
    +                                reason='insufficient-storage-for-quota')
    
    1281
    +
    
    1282
    +        elif cache_quota > cache_size + available_space:
    
    1283
    +            # The quota does not fit in the available space, this is a warning
    
    1284
    +            if '%' in self._config_cache_quota_string:
    
    1285
    +                available = (available_space / total_size) * 100
    
    1286
    +                available = '{}% of total disk space'.format(round(available, 1))
    
    1287
    +            else:
    
    1288
    +                available = utils._pretty_size(available_space)
    
    1289
    +
    
    1290
    +            self._message(Message(
    
    1291
    +                None,
    
    1292
    +                MessageType.WARN,
    
    1293
    +                "Your system does not have enough available " +
    
    1294
    +                "space to support the cache quota specified.",
    
    1295
    +                detail=("You have specified a quota of {quota} total disk space.\n" +
    
    1296
    +                        "The filesystem containing {local_cache_path} only " +
    
    1297
    +                        "has {available_size} available.")
    
    1298
    +                .format(quota=self._config_cache_quota,
    
    1299
    +                        local_cache_path=self.casdir,
    
    1300
    +                        available_size=available)))
    
    1301
    +
    
    1302
    +        # Place a slight headroom (2e9 (2GB) on the cache_quota) into
    
    1303
    +        # cache_quota to try and avoid exceptions.
    
    1304
    +        #
    
    1305
    +        # Of course, we might still end up running out during a build
    
    1306
    +        # if we end up writing more than 2G, but hey, this stuff is
    
    1307
    +        # already really fuzzy.
    
    1308
    +        #
    
    1309
    +        self._cache_quota_original = cache_quota
    
    1310
    +        self._cache_quota = cache_quota - self._cache_quota_headroom
    
    1311
    +        self._cache_lower_threshold = self._cache_quota / 2
    
    1312
    +
    
    1313
    +
    
    1046 1314
     def _grouper(iterable, n):
    
    1047 1315
         while True:
    
    1048 1316
             try:
    

  • buildstream/_context.py
    ... ... @@ -30,8 +30,8 @@ from . import _yaml
    30 30
     from ._exceptions import LoadError, LoadErrorReason, BstError
    
    31 31
     from ._message import Message, MessageType
    
    32 32
     from ._profile import Topics, profile_start, profile_end
    
    33
    -from ._artifactcache import ArtifactCache, ArtifactCacheUsage
    
    34
    -from ._cas import CASCache
    
    33
    +from ._artifactcache import ArtifactCache
    
    34
    +from ._cas import CASCache, CASQuota, CASCacheUsage
    
    35 35
     from ._workspaces import Workspaces, WorkspaceProjectCache
    
    36 36
     from .plugin import _plugin_lookup
    
    37 37
     from .sandbox import SandboxRemote
    
    ... ... @@ -155,6 +155,7 @@ class Context():
    155 155
             self._log_handle = None
    
    156 156
             self._log_filename = None
    
    157 157
             self._cascache = None
    
    158
    +        self._casquota = None
    
    158 159
             self._directory = directory
    
    159 160
     
    
    160 161
         # load()
    
    ... ... @@ -236,7 +237,15 @@ class Context():
    236 237
             cache = _yaml.node_get(defaults, Mapping, 'cache')
    
    237 238
             _yaml.node_validate(cache, ['quota', 'pull-buildtrees', 'cache-buildtrees'])
    
    238 239
     
    
    239
    -        self.config_cache_quota = _yaml.node_get(cache, str, 'quota')
    
    240
    +        self.config_cache_quota_string = _yaml.node_get(cache, str, 'quota')
    
    241
    +        try:
    
    242
    +            self.config_cache_quota = utils._parse_size(self.config_cache_quota_string,
    
    243
    +                                                        self.casdir)
    
    244
    +        except utils.UtilError as e:
    
    245
    +            raise LoadError(LoadErrorReason.INVALID_DATA,
    
    246
    +                            "{}\nPlease specify the value in bytes or as a % of full disk space.\n"
    
    247
    +                            "\nValid values are, for example: 800M 10G 1T 50%\n"
    
    248
    +                            .format(str(e))) from e
    
    240 249
     
    
    241 250
             # Load artifact share configuration
    
    242 251
             self.artifact_cache_specs = ArtifactCache.specs_from_config_node(defaults)
    
    ... ... @@ -312,15 +321,15 @@ class Context():
    312 321
     
    
    313 322
             return self._artifactcache
    
    314 323
     
    
    315
    -    # get_artifact_cache_usage()
    
    324
    +    # get_cache_usage()
    
    316 325
         #
    
    317 326
         # Fetches the current usage of the artifact cache
    
    318 327
         #
    
    319 328
         # Returns:
    
    320
    -    #     (ArtifactCacheUsage): The current status
    
    329
    +    #     (CASCacheUsage): The current status
    
    321 330
         #
    
    322
    -    def get_artifact_cache_usage(self):
    
    323
    -        return ArtifactCacheUsage(self.artifactcache)
    
    331
    +    def get_cache_usage(self):
    
    332
    +        return CASCacheUsage(self.get_casquota())
    
    324 333
     
    
    325 334
         # add_project():
    
    326 335
         #
    
    ... ... @@ -693,6 +702,11 @@ class Context():
    693 702
                 self._cascache = CASCache(self.cachedir)
    
    694 703
             return self._cascache
    
    695 704
     
    
    705
    +    def get_casquota(self):
    
    706
    +        if self._casquota is None:
    
    707
    +            self._casquota = CASQuota(self)
    
    708
    +        return self._casquota
    
    709
    +
    
    696 710
     
    
    697 711
     # _node_get_option_str()
    
    698 712
     #
    

  • buildstream/_frontend/status.py
    ... ... @@ -404,7 +404,7 @@ class _StatusHeader():
    404 404
             #
    
    405 405
             #  ~~~~~~ cache: 69% ~~~~~~
    
    406 406
             #
    
    407
    -        usage = self._context.get_artifact_cache_usage()
    
    407
    +        usage = self._context.get_cache_usage()
    
    408 408
             usage_percent = '{}%'.format(usage.used_percent)
    
    409 409
     
    
    410 410
             size = 21
    

  • buildstream/_frontend/widget.py
    ... ... @@ -486,7 +486,7 @@ class LogLine(Widget):
    486 486
             values["Session Start"] = starttime.strftime('%A, %d-%m-%Y at %H:%M:%S')
    
    487 487
             values["Project"] = "{} ({})".format(project.name, project.directory)
    
    488 488
             values["Targets"] = ", ".join([t.name for t in stream.targets])
    
    489
    -        values["Cache Usage"] = "{}".format(context.get_artifact_cache_usage())
    
    489
    +        values["Cache Usage"] = "{}".format(context.get_cache_usage())
    
    490 490
             text += self._format_values(values)
    
    491 491
     
    
    492 492
             # User configurations
    

  • buildstream/_scheduler/jobs/cachesizejob.py
    ... ... @@ -25,14 +25,14 @@ class CacheSizeJob(Job):
    25 25
             self._complete_cb = complete_cb
    
    26 26
     
    
    27 27
             context = self._scheduler.context
    
    28
    -        self._artifacts = context.artifactcache
    
    28
    +        self._casquota = context.get_casquota()
    
    29 29
     
    
    30 30
         def child_process(self):
    
    31
    -        return self._artifacts.compute_cache_size()
    
    31
    +        return self._casquota.compute_cache_size()
    
    32 32
     
    
    33 33
         def parent_complete(self, status, result):
    
    34 34
             if status == JobStatus.OK:
    
    35
    -            self._artifacts.set_cache_size(result)
    
    35
    +            self._casquota.set_cache_size(result)
    
    36 36
     
    
    37 37
             if self._complete_cb:
    
    38 38
                 self._complete_cb(status, result)
    

  • buildstream/_scheduler/jobs/cleanupjob.py
    ... ... @@ -25,27 +25,27 @@ class CleanupJob(Job):
    25 25
             self._complete_cb = complete_cb
    
    26 26
     
    
    27 27
             context = self._scheduler.context
    
    28
    +        self._casquota = context.get_casquota()
    
    28 29
             self._artifacts = context.artifactcache
    
    29 30
     
    
    30 31
         def child_process(self):
    
    31 32
             def progress():
    
    32 33
                 self.send_message('update-cache-size',
    
    33
    -                              self._artifacts.get_cache_size())
    
    34
    +                              self._casquota.get_cache_size())
    
    34 35
             return self._artifacts.clean(progress)
    
    35 36
     
    
    36 37
         def handle_message(self, message_type, message):
    
    37
    -
    
    38 38
             # Update the cache size in the main process as we go,
    
    39 39
             # this provides better feedback in the UI.
    
    40 40
             if message_type == 'update-cache-size':
    
    41
    -            self._artifacts.set_cache_size(message)
    
    41
    +            self._casquota.set_cache_size(message)
    
    42 42
                 return True
    
    43 43
     
    
    44 44
             return False
    
    45 45
     
    
    46 46
         def parent_complete(self, status, result):
    
    47 47
             if status == JobStatus.OK:
    
    48
    -            self._artifacts.set_cache_size(result)
    
    48
    +            self._casquota.set_cache_size(result)
    
    49 49
     
    
    50 50
             if self._complete_cb:
    
    51 51
                 self._complete_cb(status, result)

  • tests/artifactcache/cache_size.py
    ... ... @@ -81,11 +81,11 @@ def test_quota_over_1024T(cli, tmpdir):
    81 81
         _yaml.dump({'name': 'main'}, str(project.join("project.conf")))
    
    82 82
     
    
    83 83
         volume_space_patch = mock.patch(
    
    84
    -        "buildstream._artifactcache.ArtifactCache._get_cache_volume_size",
    
    84
    +        "buildstream._cas.CASQuota._get_cache_volume_size",
    
    85 85
             autospec=True,
    
    86 86
             return_value=(1025 * TiB, 1025 * TiB)
    
    87 87
         )
    
    88 88
     
    
    89 89
         with volume_space_patch:
    
    90 90
             result = cli.run(project, args=["build", "file.bst"])
    
    91
    -        result.assert_main_error(ErrorDomain.ARTIFACT, 'insufficient-storage-for-quota')
    91
    +        result.assert_main_error(ErrorDomain.CAS, 'insufficient-storage-for-quota')

  • tests/artifactcache/expiry.py
    ... ... @@ -341,7 +341,7 @@ def test_never_delete_required_track(cli, datafiles, tmpdir):
    341 341
         ("200%", ErrorDomain.LOAD, LoadErrorReason.INVALID_DATA),
    
    342 342
     
    
    343 343
         # Not enough space on disk even if you cleaned up
    
    344
    -    ("11K", ErrorDomain.ARTIFACT, 'insufficient-storage-for-quota'),
    
    344
    +    ("11K", ErrorDomain.CAS, 'insufficient-storage-for-quota'),
    
    345 345
     
    
    346 346
         # Not enough space for these caches
    
    347 347
         ("7K", 'warning', 'Your system does not have enough available'),
    
    ... ... @@ -355,7 +355,7 @@ def test_invalid_cache_quota(cli, datafiles, tmpdir, quota, err_domain, err_reas
    355 355
         cli.configure({
    
    356 356
             'cache': {
    
    357 357
                 'quota': quota,
    
    358
    -        }
    
    358
    +        },
    
    359 359
         })
    
    360 360
     
    
    361 361
         # We patch how we get space information
    
    ... ... @@ -373,13 +373,13 @@ def test_invalid_cache_quota(cli, datafiles, tmpdir, quota, err_domain, err_reas
    373 373
             total_space = 10000
    
    374 374
     
    
    375 375
         volume_space_patch = mock.patch(
    
    376
    -        "buildstream._artifactcache.ArtifactCache._get_cache_volume_size",
    
    376
    +        "buildstream.utils._get_volume_size",
    
    377 377
             autospec=True,
    
    378 378
             return_value=(total_space, free_space),
    
    379 379
         )
    
    380 380
     
    
    381 381
         cache_size_patch = mock.patch(
    
    382
    -        "buildstream._artifactcache.ArtifactCache.get_cache_size",
    
    382
    +        "buildstream._cas.CASQuota.get_cache_size",
    
    383 383
             autospec=True,
    
    384 384
             return_value=0,
    
    385 385
         )
    

  • tests/testutils/artifactshare.py
    ... ... @@ -46,7 +46,6 @@ class ArtifactShare():
    46 46
             # in tests as a remote artifact push/pull configuration
    
    47 47
             #
    
    48 48
             self.repodir = os.path.join(self.directory, 'repo')
    
    49
    -
    
    50 49
             os.makedirs(self.repodir)
    
    51 50
     
    
    52 51
             self.cas = CASCache(self.repodir)
    



  • [Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]