1 # -*- encoding: binary -*-
2 # here are internal implementation details, do not use them in your code
6 require 'mogilefs/chunker'
9 # HTTPFile wraps up the new file operations for storing files onto an HTTP
12 # You really don't want to create an HTTPFile by hand. Instead you want to
13 # create a new file using MogileFS::MogileFS.new_file.
15 class MogileFS::HTTPFile < StringIO
16 class RetryableError < MogileFS::Error; end
17 class EmptyResponseError < RetryableError; end
18 class BadResponseError < RetryableError; end
19 class UnparseableResponseError < RetryableError; end
20 class NoStorageNodesError < MogileFS::Error
21 def message; 'Unable to open socket to storage node'; end
23 class NonRetryableError < MogileFS::Error; end
25 class HTTPSock < MogileFS::Socket
28 # Increase timeout as we become more invested in uploading with
29 # this socket. The server could be experiencing I/O delays
30 # from large uploads because the sysadmin forgot to tune the
31 # VM sysctls for handling large files.
33 timed_write(buf, Time.now - @start + 5.0)
38 MD5_TRAILER_NODES = {} # :nodoc: # EXPERIMENTAL
40 attr_accessor :response_timeout_cb
43 # temporary directories (nginx) may not be configured on the
44 # same device, necessitating a time-consuming full file copy
45 # instead of a quick rename(2)/link(2) operation
46 @response_timeout_cb = lambda do |elapsed_time, bytes_uploaded|
47 mbytes_uploaded = bytes_uploaded / (1024.0 * 1024.0)
48 # assumes worst case is 10M/s on the remote storage disk
49 t = mbytes_uploaded * 10 + elapsed_time
55 # The big_io name in case we have file > 256M
59 attr_accessor :streaming_io
62 # Creates a new HTTPFile with MogileFS-specific data. Use
63 # MogileFS::MogileFS#new_file instead of this method.
65 def initialize(dests, opts = nil)
67 @md5 = @streaming_io = @big_io = @active = nil
69 @opts = Integer === opts ? { :content_length => opts } : opts
72 def request_put(sock, uri, file_size, input = nil)
73 host_with_port = "#{uri.host}:#{uri.port}"
74 clen = @opts[:content_length]
77 content_md5 = @opts[:content_md5]
78 if String === content_md5
81 ":content_length must be specified with :content_md5 String"
82 file_size = "#{file_size}\r\nContent-MD5: #{content_md5}"
83 elsif content_md5.respond_to?(:call) ||
84 :trailer == content_md5 ||
85 MD5_TRAILER_NODES[host_with_port]
87 @md5 = Digest::MD5.new
91 sock.write("PUT #{uri.request_uri} HTTP/1.0\r\n" \
92 "Content-Length: #{file_size}\r\n\r\n")
93 rv = input ? MogileFS.io.copy_stream(@active = input, sock) : yield(sock)
95 trailers = @md5 ? "Trailer: Content-MD5\r\n" : ""
96 sock.write("PUT #{uri.request_uri} HTTP/1.1\r\n" \
97 "Host: #{host_with_port}\r\n#{trailers}" \
98 "Transfer-Encoding: chunked\r\n\r\n")
99 tmp = MogileFS::Chunker.new(sock, @md5, content_md5)
100 rv = input ? MogileFS.io.copy_stream(@active = input, tmp) : yield(tmp)
104 if clen && clen != rv
105 raise MogileFS::SizeMismatchError,
106 ":content_length expected: #{clen.inspect}, actual: #{rv.inspect}"
111 def put_streaming_io(sock, uri) # unlikely to be used
112 file_size = @streaming_io.length
114 request_put(sock, uri, file_size) do |wr|
115 @streaming_io.call(Proc.new do |data_to_write|
116 written += wr.write(data_to_write)
119 file_size ? file_size : written
122 def rewind_or_raise!(uri, err)
123 @active.rewind if @active
125 msg = "#{uri} failed with #{err.message} (#{err.class}) and " \
126 "retrying is impossible as rewind on " \
127 "#{@active.inspect} failed with: #{e.message} (#{e.class})"
128 raise NonRetryableError, msg, e.backtrace
132 # Writes an HTTP PUT request to +sock+ to upload the file and
133 # returns file size if the socket finished writing
134 def upload(devid, uri) # :nodoc:
136 sock = HTTPSock.tcp(uri.host, uri.port)
141 file_size = put_streaming_io(sock, uri)
143 if String === @big_io || @big_io.respond_to?(:to_path)
144 file = File.open(@big_io)
146 file_size = request_put(sock, uri, stat.file? ? stat.size : nil, file)
149 if @big_io.respond_to?(:stat)
151 size = stat.size if stat.file?
152 elsif @big_io.respond_to?(:size)
155 file_size = request_put(sock, uri, size, @big_io)
159 request_put(sock, uri, file_size, self)
162 tout = self.class.response_timeout_cb.call(Time.now - start, file_size)
164 case line = sock.timed_read(23, "", tout)
165 when %r{^HTTP/\d\.\d\s+(2\d\d)\s} # success!
168 raise EmptyResponseError, 'Unable to read response line from server'
169 when %r{^HTTP/\d\.\d\s+(\d+)}
170 raise BadResponseError, "HTTP response status from upload: #$1"
172 raise UnparseableResponseError,
173 "Response line not understood: #{line.inspect}"
175 rescue SystemCallError, RetryableError => err
176 rewind_or_raise!(uri, err)
185 @dests.each do |devid, path|
187 uri = URI.parse(path)
188 bytes_uploaded = upload(devid, uri)
189 return create_close(devid, uri, bytes_uploaded)
190 rescue SystemCallError, RetryableError => e
192 errors << "#{path} - #{e.message} (#{e.class})"
196 raise NoStorageNodesError,
197 "all paths failed with PUT: #{errors.join(', ')}", []
200 def create_close(devid, uri, bytes_uploaded)
205 :domain => @opts[:domain],
206 :size => bytes_uploaded,
210 args[:checksum] = "MD5:#{@md5.hexdigest}"
211 elsif String === @opts[:content_md5]
212 hex = @opts[:content_md5].unpack('m')[0].unpack('H*')[0]
213 args[:checksum] = "MD5:#{hex}"
215 args[:checksumverify] = 1 if @opts[:checksumverify]
216 @opts[:backend].create_close(args)