##// END OF EJS Templates
compression: introduce a `storage.revlog.zlib.level` configuration...
compression: introduce a `storage.revlog.zlib.level` configuration This option control the zlib compression level used when compression revlog chunk. This is also a good excuse to pave the way for a similar configuration option for the zstd compression engine. Having a dedicated option for each compression algorithm is useful because they don't support the same range of values. Using a higher zlib compression impact CPU consumption at compression time, but does not directly affected decompression time. However dealing with small compressed chunk can directly help decompression and indirectly help other revlog logic. I ran some basic test on repositories using different level. I am using the mercurial, pypy, netbeans and mozilla-central clone from our benchmark suite. All tested repository use sparse-revlog and got all their delta recomputed. The different compression level has a small effect on the repository size (about 10% variation in the total range). My quick analysis is that revlog mostly store small delta, that are not affected by the compression level much. So the variation probably mostly comes from better compression of the snapshots revisions, and snapshot revision only represent a small portion of the repository content. I also made some basic timings measurements. The "read" timings are gathered using simple run of `hg perfrevlogrevisions`, the "write" timings using `hg perfrevlogwrite` (restricted to the last 5000 revisions for netbeans and mozilla central). The timings are gathered on a generic machine, (not one of our performance locked machine), so small variation might not be meaningful. However large trend remains relevant. Keep in mind that these numbers are not pure compression/decompression time. They also involve the full revlog logic. In particular the difference in chunk size has an impact on the delta chain structure, affecting performance when writing or reading them. On read/write performance, the compression level has a bigger impact. Counter-intuitively, the higher compression levels improve "write" performance for the large repositories in our tested setting. Maybe because the last 5000 delta chain end up having a very different shape in this specific spot? Or maybe because of a more general trend of better delta chains thanks to the smaller chunk and snapshot. This series does not intend to change the default compression level. However, these result call for a deeper analysis of this performance difference in the future. Full data ========= repo level .hg/store size 00manifest.d read write ---------------------------------------------------------------- mercurial 1 49,402,813 5,963,475 0.170159 53.250304 mercurial 6 47,197,397 5,875,730 0.182820 56.264320 mercurial 9 47,121,596 5,849,781 0.189219 56.293612 pypy 1 370,830,572 28,462,425 2.679217 460.721984 pypy 6 340,112,317 27,648,747 2.768691 467.537158 pypy 9 338,360,736 27,639,003 2.763495 476.589918 netbeans 1 1,281,847,810 165,495,457 122.477027 520.560316 netbeans 6 1,205,284,353 159,161,207 139.876147 715.930400 netbeans 9 1,197,135,671 155,034,586 141.620281 678.297064 mozilla 1 2,775,497,186 298,527,987 147.867662 751.263721 mozilla 6 2,596,856,420 286,597,671 170.572118 987.056093 mozilla 9 2,587,542,494 287,018,264 163.622338 739.803002

File last commit:

r41470:6d7f18cd default
r42210:1fac9b93 default
Show More
wireprotolfsserver.py
352 lines | 11.7 KiB | text/x-python | PythonLexer
/ hgext / lfs / wireprotolfsserver.py
# wireprotolfsserver.py - lfs protocol server side implementation
#
# Copyright 2018 Matt Harbison <matt_harbison@yahoo.com>
#
# This software may be used and distributed according to the terms of the
# GNU General Public License version 2 or any later version.
from __future__ import absolute_import
import datetime
import errno
import json
import traceback
from mercurial.hgweb import (
common as hgwebcommon,
)
from mercurial import (
exthelper,
pycompat,
util,
wireprotoserver,
)
from . import blobstore
HTTP_OK = hgwebcommon.HTTP_OK
HTTP_CREATED = hgwebcommon.HTTP_CREATED
HTTP_BAD_REQUEST = hgwebcommon.HTTP_BAD_REQUEST
HTTP_NOT_FOUND = hgwebcommon.HTTP_NOT_FOUND
HTTP_METHOD_NOT_ALLOWED = hgwebcommon.HTTP_METHOD_NOT_ALLOWED
HTTP_NOT_ACCEPTABLE = hgwebcommon.HTTP_NOT_ACCEPTABLE
HTTP_UNSUPPORTED_MEDIA_TYPE = hgwebcommon.HTTP_UNSUPPORTED_MEDIA_TYPE
eh = exthelper.exthelper()
@eh.wrapfunction(wireprotoserver, 'handlewsgirequest')
def handlewsgirequest(orig, rctx, req, res, checkperm):
"""Wrap wireprotoserver.handlewsgirequest() to possibly process an LFS
request if it is left unprocessed by the wrapped method.
"""
if orig(rctx, req, res, checkperm):
return True
if not rctx.repo.ui.configbool(b'experimental', b'lfs.serve'):
return False
if not util.safehasattr(rctx.repo.svfs, 'lfslocalblobstore'):
return False
if not req.dispatchpath:
return False
try:
if req.dispatchpath == b'.git/info/lfs/objects/batch':
checkperm(rctx, req, b'pull')
return _processbatchrequest(rctx.repo, req, res)
# TODO: reserve and use a path in the proposed http wireprotocol /api/
# namespace?
elif req.dispatchpath.startswith(b'.hg/lfs/objects'):
return _processbasictransfer(rctx.repo, req, res,
lambda perm:
checkperm(rctx, req, perm))
return False
except hgwebcommon.ErrorResponse as e:
# XXX: copied from the handler surrounding wireprotoserver._callhttp()
# in the wrapped function. Should this be moved back to hgweb to
# be a common handler?
for k, v in e.headers:
res.headers[k] = v
res.status = hgwebcommon.statusmessage(e.code, pycompat.bytestr(e))
res.setbodybytes(b'0\n%s\n' % pycompat.bytestr(e))
return True
def _sethttperror(res, code, message=None):
res.status = hgwebcommon.statusmessage(code, message=message)
res.headers[b'Content-Type'] = b'text/plain; charset=utf-8'
res.setbodybytes(b'')
def _logexception(req):
"""Write information about the current exception to wsgi.errors."""
tb = pycompat.sysbytes(traceback.format_exc())
errorlog = req.rawenv[b'wsgi.errors']
uri = b''
if req.apppath:
uri += req.apppath
uri += b'/' + req.dispatchpath
errorlog.write(b"Exception happened while processing request '%s':\n%s" %
(uri, tb))
def _processbatchrequest(repo, req, res):
"""Handle a request for the Batch API, which is the gateway to granting file
access.
https://github.com/git-lfs/git-lfs/blob/master/docs/api/batch.md
"""
# Mercurial client request:
#
# HOST: localhost:$HGPORT
# ACCEPT: application/vnd.git-lfs+json
# ACCEPT-ENCODING: identity
# USER-AGENT: git-lfs/2.3.4 (Mercurial 4.5.2+1114-f48b9754f04c+20180316)
# Content-Length: 125
# Content-Type: application/vnd.git-lfs+json
#
# {
# "objects": [
# {
# "oid": "31cf...8e5b"
# "size": 12
# }
# ]
# "operation": "upload"
# }
if req.method != b'POST':
_sethttperror(res, HTTP_METHOD_NOT_ALLOWED)
return True
if req.headers[b'Content-Type'] != b'application/vnd.git-lfs+json':
_sethttperror(res, HTTP_UNSUPPORTED_MEDIA_TYPE)
return True
if req.headers[b'Accept'] != b'application/vnd.git-lfs+json':
_sethttperror(res, HTTP_NOT_ACCEPTABLE)
return True
# XXX: specify an encoding?
lfsreq = json.loads(req.bodyfh.read())
# If no transfer handlers are explicitly requested, 'basic' is assumed.
if r'basic' not in lfsreq.get(r'transfers', [r'basic']):
_sethttperror(res, HTTP_BAD_REQUEST,
b'Only the basic LFS transfer handler is supported')
return True
operation = lfsreq.get(r'operation')
operation = pycompat.bytestr(operation)
if operation not in (b'upload', b'download'):
_sethttperror(res, HTTP_BAD_REQUEST,
b'Unsupported LFS transfer operation: %s' % operation)
return True
localstore = repo.svfs.lfslocalblobstore
objects = [p for p in _batchresponseobjects(req, lfsreq.get(r'objects', []),
operation, localstore)]
rsp = {
r'transfer': r'basic',
r'objects': objects,
}
res.status = hgwebcommon.statusmessage(HTTP_OK)
res.headers[b'Content-Type'] = b'application/vnd.git-lfs+json'
res.setbodybytes(pycompat.bytestr(json.dumps(rsp)))
return True
def _batchresponseobjects(req, objects, action, store):
"""Yield one dictionary of attributes for the Batch API response for each
object in the list.
req: The parsedrequest for the Batch API request
objects: The list of objects in the Batch API object request list
action: 'upload' or 'download'
store: The local blob store for servicing requests"""
# Successful lfs-test-server response to solict an upload:
# {
# u'objects': [{
# u'size': 12,
# u'oid': u'31cf...8e5b',
# u'actions': {
# u'upload': {
# u'href': u'http://localhost:$HGPORT/objects/31cf...8e5b',
# u'expires_at': u'0001-01-01T00:00:00Z',
# u'header': {
# u'Accept': u'application/vnd.git-lfs'
# }
# }
# }
# }]
# }
# TODO: Sort out the expires_at/expires_in/authenticated keys.
for obj in objects:
# Convert unicode to ASCII to create a filesystem path
soid = obj.get(r'oid')
oid = soid.encode(r'ascii')
rsp = {
r'oid': soid,
r'size': obj.get(r'size'), # XXX: should this check the local size?
#r'authenticated': True,
}
exists = True
verifies = False
# Verify an existing file on the upload request, so that the client is
# solicited to re-upload if it corrupt locally. Download requests are
# also verified, so the error can be flagged in the Batch API response.
# (Maybe we can use this to short circuit the download for `hg verify`,
# IFF the client can assert that the remote end is an hg server.)
# Otherwise, it's potentially overkill on download, since it is also
# verified as the file is streamed to the caller.
try:
verifies = store.verify(oid)
if verifies and action == b'upload':
# The client will skip this upload, but make sure it remains
# available locally.
store.linkfromusercache(oid)
except IOError as inst:
if inst.errno != errno.ENOENT:
_logexception(req)
rsp[r'error'] = {
r'code': 500,
r'message': inst.strerror or r'Internal Server Server'
}
yield rsp
continue
exists = False
# Items are always listed for downloads. They are dropped for uploads
# IFF they already exist locally.
if action == b'download':
if not exists:
rsp[r'error'] = {
r'code': 404,
r'message': r"The object does not exist"
}
yield rsp
continue
elif not verifies:
rsp[r'error'] = {
r'code': 422, # XXX: is this the right code?
r'message': r"The object is corrupt"
}
yield rsp
continue
elif verifies:
yield rsp # Skip 'actions': already uploaded
continue
expiresat = datetime.datetime.now() + datetime.timedelta(minutes=10)
def _buildheader():
# The spec doesn't mention the Accept header here, but avoid
# a gratuitous deviation from lfs-test-server in the test
# output.
hdr = {
r'Accept': r'application/vnd.git-lfs'
}
auth = req.headers.get(b'Authorization', b'')
if auth.startswith(b'Basic '):
hdr[r'Authorization'] = pycompat.strurl(auth)
return hdr
rsp[r'actions'] = {
r'%s' % pycompat.strurl(action): {
r'href': pycompat.strurl(b'%s%s/.hg/lfs/objects/%s'
% (req.baseurl, req.apppath, oid)),
# datetime.isoformat() doesn't include the 'Z' suffix
r"expires_at": expiresat.strftime(r'%Y-%m-%dT%H:%M:%SZ'),
r'header': _buildheader(),
}
}
yield rsp
def _processbasictransfer(repo, req, res, checkperm):
"""Handle a single file upload (PUT) or download (GET) action for the Basic
Transfer Adapter.
After determining if the request is for an upload or download, the access
must be checked by calling ``checkperm()`` with either 'pull' or 'upload'
before accessing the files.
https://github.com/git-lfs/git-lfs/blob/master/docs/api/basic-transfers.md
"""
method = req.method
oid = req.dispatchparts[-1]
localstore = repo.svfs.lfslocalblobstore
if len(req.dispatchparts) != 4:
_sethttperror(res, HTTP_NOT_FOUND)
return True
if method == b'PUT':
checkperm(b'upload')
# TODO: verify Content-Type?
existed = localstore.has(oid)
# TODO: how to handle timeouts? The body proxy handles limiting to
# Content-Length, but what happens if a client sends less than it
# says it will?
statusmessage = hgwebcommon.statusmessage
try:
localstore.download(oid, req.bodyfh)
res.status = statusmessage(HTTP_OK if existed else HTTP_CREATED)
except blobstore.LfsCorruptionError:
_logexception(req)
# XXX: Is this the right code?
res.status = statusmessage(422, b'corrupt blob')
# There's no payload here, but this is the header that lfs-test-server
# sends back. This eliminates some gratuitous test output conditionals.
res.headers[b'Content-Type'] = b'text/plain; charset=utf-8'
res.setbodybytes(b'')
return True
elif method == b'GET':
checkperm(b'pull')
res.status = hgwebcommon.statusmessage(HTTP_OK)
res.headers[b'Content-Type'] = b'application/octet-stream'
try:
# TODO: figure out how to send back the file in chunks, instead of
# reading the whole thing. (Also figure out how to send back
# an error status if an IOError occurs after a partial write
# in that case. Here, everything is read before starting.)
res.setbodybytes(localstore.read(oid))
except blobstore.LfsCorruptionError:
_logexception(req)
# XXX: Is this the right code?
res.status = hgwebcommon.statusmessage(422, b'corrupt blob')
res.setbodybytes(b'')
return True
else:
_sethttperror(res, HTTP_METHOD_NOT_ALLOWED,
message=b'Unsupported LFS transfer method: %s' % method)
return True