backupii 0.1.0.pre.alpha.1

data/lib/backup/cloud_io/cloud_files.rb

@@ -0,0 +1,301 @@
+# frozen_string_literal: true
+
+require "backup/cloud_io/base"
+require "fog"
+require "digest/md5"
+
+module Backup
+  module CloudIO
+    class CloudFiles < Base
+      class Error < Backup::Error; end
+
+      MAX_FILE_SIZE   = 1024**3 * 5     # 5 GiB
+      MAX_SLO_SIZE    = 1024**3 * 5000  # 1000 segments @ 5 GiB
+      SEGMENT_BUFFER  = 1024**2         # 1 MiB
+
+      attr_reader :username, :api_key, :auth_url, :region, :servicenet,
+        :container, :segments_container, :segment_size, :days_to_keep,
+        :fog_options
+
+      def initialize(options = {})
+        super
+
+        @username           = options[:username]
+        @api_key            = options[:api_key]
+        @auth_url           = options[:auth_url]
+        @region             = options[:region]
+        @servicenet         = options[:servicenet]
+        @container          = options[:container]
+        @segments_container = options[:segments_container]
+        @segment_size       = options[:segment_size]
+        @days_to_keep       = options[:days_to_keep]
+        @fog_options        = options[:fog_options]
+      end
+
+      # The Syncer may call this method in multiple threads,
+      # but #objects is always called before this occurs.
+      def upload(src, dest)
+        create_containers
+
+        file_size = File.size(src)
+        segment_bytes = segment_size * 1024**2
+        if segment_bytes > 0 && file_size > segment_bytes
+          raise FileSizeError, <<-EOS if file_size > MAX_SLO_SIZE
+            File Too Large
+            File: #{src}
+            Size: #{file_size}
+            Max SLO Size is #{MAX_SLO_SIZE} (5 GiB * 1000 segments)
+          EOS
+
+          segment_bytes = adjusted_segment_bytes(segment_bytes, file_size)
+          segments = upload_segments(src, dest, segment_bytes, file_size)
+          upload_manifest(dest, segments)
+        else
+          raise FileSizeError, <<-EOS if file_size > MAX_FILE_SIZE
+            File Too Large
+            File: #{src}
+            Size: #{file_size}
+            Max File Size is #{MAX_FILE_SIZE} (5 GiB)
+          EOS
+
+          put_object(src, dest)
+        end
+      end
+
+      # Returns all objects in the container with the given prefix.
+      #
+      # - #get_container returns a max of 10000 objects per request.
+      # - Returns objects sorted using a sqlite binary collating function.
+      # - If marker is given, only objects after the marker are in the response.
+      def objects(prefix)
+        objects = []
+        resp = nil
+        prefix = prefix.chomp("/")
+        opts = { prefix: prefix + "/" }
+
+        create_containers
+
+        while resp.nil? || resp.body.count == 10_000
+          opts[:marker] = objects.last.name unless objects.empty?
+          with_retries("GET '#{container}/#{prefix}/*'") do
+            resp = connection.get_container(container, opts)
+          end
+          resp.body.each do |obj_data|
+            objects << Object.new(self, obj_data)
+          end
+        end
+
+        objects
+      end
+
+      # Used by Object to fetch metadata if needed.
+      def head_object(object)
+        resp = nil
+        with_retries("HEAD '#{container}/#{object.name}'") do
+          resp = connection.head_object(container, object.name)
+        end
+        resp
+      end
+
+      # Delete non-SLO object(s) from the container.
+      #
+      # - Called by the Storage (with objects) and the Syncer (with names)
+      # - Deletes 10,000 objects per request.
+      # - Missing objects will be ignored.
+      def delete(objects_or_names)
+        names = Array(objects_or_names).dup
+        names.map!(&:name) if names.first.is_a?(Object)
+
+        until names.empty?
+          names_partial = names.slice!(0, 10_000)
+          with_retries("DELETE Multiple Objects") do
+            resp = connection.delete_multiple_objects(container, names_partial)
+            resp_status = resp.body["Response Status"]
+            raise Error, <<-EOS unless resp_status == "200 OK"
+              #{resp_status}
+              The server returned the following:
+              #{resp.body.inspect}
+            EOS
+          end
+        end
+      end
+
+      # Delete an SLO object(s) from the container.
+      #
+      # - Used only by the Storage. The Syncer cannot use SLOs.
+      # - Removes the SLO manifest object and all associated segments.
+      # - Missing segments will be ignored.
+      def delete_slo(objects)
+        Array(objects).each do |object|
+          with_retries("DELETE SLO Manifest '#{container}/#{object.name}'") do
+            resp = connection.delete_static_large_object(container, object.name)
+            resp_status = resp.body["Response Status"]
+            raise Error, <<-EOS unless resp_status == "200 OK"
+              #{resp_status}
+              The server returned the following:
+              #{resp.body.inspect}
+            EOS
+          end
+        end
+      end
+
+      private
+
+      def connection
+        @connection ||= Fog::Storage.new({
+          provider: "Rackspace",
+          rackspace_username: username,
+          rackspace_api_key: api_key,
+          rackspace_auth_url: auth_url,
+          rackspace_region: region,
+          rackspace_servicenet: servicenet
+        }.merge(fog_options || {}))
+      end
+
+      def create_containers
+        return if @containers_created
+
+        @containers_created = true
+
+        with_retries("Create Containers") do
+          connection.put_container(container)
+          connection.put_container(segments_container) if segments_container
+        end
+      end
+
+      def put_object(src, dest)
+        opts = headers.merge("ETag" => Digest::MD5.file(src).hexdigest)
+        with_retries("PUT '#{container}/#{dest}'") do
+          File.open(src, "r") do |file|
+            connection.put_object(container, dest, file, opts)
+          end
+        end
+      end
+
+      # Each segment is uploaded using chunked transfer encoding using
+      # SEGMENT_BUFFER, and each segment's MD5 is sent to verify the transfer.
+      # Each segment's MD5 and byte_size will also be verified when the
+      # SLO manifest object is uploaded.
+      def upload_segments(src, dest, segment_bytes, file_size)
+        total_segments = (file_size / segment_bytes.to_f).ceil
+        progress = (0.1..0.9).step(0.1).map { |n| (total_segments * n).floor }
+        Logger.info "\s\sUploading #{total_segments} SLO Segments..."
+
+        segments = []
+        File.open(src, "r") do |file|
+          segment_number = 0
+          until file.eof?
+            segment_number += 1
+            object = "#{dest}/#{segment_number.to_s.rjust(4, "0")}"
+            pos = file.pos
+            md5 = segment_md5(file, segment_bytes)
+            opts = headers.merge("ETag" => md5)
+
+            with_retries("PUT '#{segments_container}/#{object}'") do
+              file.seek(pos)
+              offset = 0
+              connection.put_object(segments_container, object, nil, opts) do
+                # block is called to stream data until it returns ''
+                data = ""
+                if offset <= segment_bytes - SEGMENT_BUFFER
+                  data = file.read(SEGMENT_BUFFER).to_s # nil => ''
+                  offset += data.size
+                end
+                data
+              end
+            end
+
+            segments << {
+              path: "#{segments_container}/#{object}",
+              etag: md5,
+              size_bytes: file.pos - pos
+            }
+
+            if (i = progress.rindex(segment_number))
+              Logger.info "\s\s...#{i + 1}0% Complete..."
+            end
+          end
+        end
+        segments
+      end
+
+      def segment_md5(file, segment_bytes)
+        md5 = Digest::MD5.new
+        offset = 0
+        while offset <= segment_bytes - SEGMENT_BUFFER
+          data = file.read(SEGMENT_BUFFER)
+          break unless data
+
+          offset += data.size
+          md5 << data
+        end
+        md5.hexdigest
+      end
+
+      # Each segment's ETag and byte_size will be verified once uploaded.
+      # Request will raise an exception if verification fails or segments
+      # are not found. However, each segment's ETag was verified when we
+      # uploaded the segments, so this should only retry failed requests.
+      def upload_manifest(dest, segments)
+        Logger.info "\s\sStoring SLO Manifest '#{container}/#{dest}'"
+
+        with_retries("PUT SLO Manifest '#{container}/#{dest}'") do
+          connection.put_static_obj_manifest(container, dest, segments, headers)
+        end
+      end
+
+      # If :days_to_keep was set, each object will be scheduled for deletion.
+      # This includes non-SLO objects, the SLO manifest and all segments.
+      def headers
+        headers = {}
+        headers["X-Delete-At"] = delete_at if delete_at
+        headers
+      end
+
+      def delete_at
+        return unless days_to_keep
+
+        @delete_at ||= (Time.now.utc + days_to_keep * 60**2 * 24).to_i
+      end
+
+      def adjusted_segment_bytes(segment_bytes, file_size)
+        return segment_bytes if file_size / segment_bytes.to_f <= 1000
+
+        mb = orig_mb = segment_bytes / 1024**2
+        mb += 1 until file_size / (1024**2 * mb).to_f <= 1000
+        Logger.warn Error.new(<<-EOS)
+          Segment Size Adjusted
+          Your original #segment_size of #{orig_mb} MiB has been adjusted
+          to #{mb} MiB in order to satisfy the limit of 1000 segments.
+          To enforce your chosen #segment_size, you should use the Splitter.
+          e.g. split_into_chunks_of #{mb * 1000} (#segment_size * 1000)
+        EOS
+        1024**2 * mb
+      end
+
+      class Object
+        attr_reader :name, :hash
+
+        def initialize(cloud_io, data)
+          @cloud_io = cloud_io
+          @name = data["name"]
+          @hash = data["hash"]
+        end
+
+        def slo?
+          !!metadata["X-Static-Large-Object"]
+        end
+
+        def marked_for_deletion?
+          !!metadata["X-Delete-At"]
+        end
+
+        private
+
+        def metadata
+          @metadata ||= @cloud_io.head_object(self).headers
+        end
+      end
+    end
+  end
+end

data/lib/backup/cloud_io/s3.rb

@@ -0,0 +1,256 @@
+# frozen_string_literal: true
+
+require "backup/cloud_io/base"
+require "fog"
+require "digest/md5"
+require "base64"
+require "stringio"
+
+module Backup
+  module CloudIO
+    class S3 < Base
+      class Error < Backup::Error; end
+
+      MAX_FILE_SIZE       = 1024**3 * 5   # 5 GiB
+      MAX_MULTIPART_SIZE  = 1024**4 * 5   # 5 TiB
+
+      attr_reader :access_key_id, :secret_access_key, :use_iam_profile,
+        :region, :bucket, :chunk_size, :encryption, :storage_class,
+        :fog_options
+
+      def initialize(options = {})
+        super
+
+        @access_key_id      = options[:access_key_id]
+        @secret_access_key  = options[:secret_access_key]
+        @use_iam_profile    = options[:use_iam_profile]
+        @region             = options[:region]
+        @bucket             = options[:bucket]
+        @chunk_size         = options[:chunk_size]
+        @encryption         = options[:encryption]
+        @storage_class      = options[:storage_class]
+        @fog_options        = options[:fog_options]
+      end
+
+      # The Syncer may call this method in multiple threads.
+      # However, #objects is always called prior to multithreading.
+      def upload(src, dest)
+        file_size = File.size(src)
+        chunk_bytes = chunk_size * 1024**2
+        if chunk_bytes > 0 && file_size > chunk_bytes
+          raise FileSizeError, <<-EOS if file_size > MAX_MULTIPART_SIZE
+            File Too Large
+            File: #{src}
+            Size: #{file_size}
+            Max Multipart Upload Size is #{MAX_MULTIPART_SIZE} (5 TiB)
+          EOS
+
+          chunk_bytes = adjusted_chunk_bytes(chunk_bytes, file_size)
+          upload_id = initiate_multipart(dest)
+          parts = upload_parts(src, dest, upload_id, chunk_bytes, file_size)
+          complete_multipart(dest, upload_id, parts)
+        else
+          raise FileSizeError, <<-EOS if file_size > MAX_FILE_SIZE
+            File Too Large
+            File: #{src}
+            Size: #{file_size}
+            Max File Size is #{MAX_FILE_SIZE} (5 GiB)
+          EOS
+
+          put_object(src, dest)
+        end
+      end
+
+      # Returns all objects in the bucket with the given prefix.
+      #
+      # - #get_bucket returns a max of 1000 objects per request.
+      # - Returns objects in alphabetical order.
+      # - If marker is given, only objects after the marker are in the response.
+      def objects(prefix)
+        objects = []
+        resp = nil
+        prefix = prefix.chomp("/")
+        opts = { "prefix" => prefix + "/" }
+
+        while resp.nil? || resp.body["IsTruncated"]
+          opts["marker"] = objects.last.key unless objects.empty?
+          with_retries("GET '#{bucket}/#{prefix}/*'") do
+            resp = connection.get_bucket(bucket, opts)
+          end
+          resp.body["Contents"].each do |obj_data|
+            objects << Object.new(self, obj_data)
+          end
+        end
+
+        objects
+      end
+
+      # Used by Object to fetch metadata if needed.
+      def head_object(object)
+        resp = nil
+        with_retries("HEAD '#{bucket}/#{object.key}'") do
+          resp = connection.head_object(bucket, object.key)
+        end
+        resp
+      end
+
+      # Delete object(s) from the bucket.
+      #
+      # - Called by the Storage (with objects) and the Syncer (with keys)
+      # - Deletes 1000 objects per request.
+      # - Missing objects will be ignored.
+      def delete(objects_or_keys)
+        keys = Array(objects_or_keys).dup
+        keys.map!(&:key) if keys.first.is_a?(Object)
+
+        opts = { quiet: true } # only report Errors in DeleteResult
+        until keys.empty?
+          keys_partial = keys.slice!(0, 1000)
+          with_retries("DELETE Multiple Objects") do
+            resp = connection.delete_multiple_objects(bucket, keys_partial, opts.dup)
+            unless resp.body["DeleteResult"].empty?
+              errors = resp.body["DeleteResult"].map do |result|
+                error = result["Error"]
+                "Failed to delete: #{error["Key"]}\n" \
+                  "Reason: #{error["Code"]}: #{error["Message"]}"
+              end.join("\n")
+              raise Error, "The server returned the following:\n#{errors}"
+            end
+          end
+        end
+      end
+
+      private
+
+      def connection
+        @connection ||=
+          begin
+            opts = { provider: "AWS", region: region }
+            if use_iam_profile
+              opts[:use_iam_profile] = true
+            else
+              opts[:aws_access_key_id] = access_key_id
+              opts[:aws_secret_access_key] = secret_access_key
+            end
+            opts.merge!(fog_options || {})
+            conn = Fog::Storage.new(opts)
+            conn.sync_clock
+            conn
+          end
+      end
+
+      def put_object(src, dest)
+        md5 = Base64.encode64(Digest::MD5.file(src).digest).chomp
+        options = headers.merge("Content-MD5" => md5)
+        with_retries("PUT '#{bucket}/#{dest}'") do
+          File.open(src, "r") do |file|
+            connection.put_object(bucket, dest, file, options)
+          end
+        end
+      end
+
+      def initiate_multipart(dest)
+        Logger.info "\s\sInitiate Multipart '#{bucket}/#{dest}'"
+
+        resp = nil
+        with_retries("POST '#{bucket}/#{dest}' (Initiate)") do
+          resp = connection.initiate_multipart_upload(bucket, dest, headers)
+        end
+        resp.body["UploadId"]
+      end
+
+      # Each part's MD5 is sent to verify the transfer.
+      # AWS will concatenate all parts into a single object
+      # once the multipart upload is completed.
+      def upload_parts(src, dest, upload_id, chunk_bytes, file_size)
+        total_parts = (file_size / chunk_bytes.to_f).ceil
+        progress = (0.1..0.9).step(0.1).map { |n| (total_parts * n).floor }
+        Logger.info "\s\sUploading #{total_parts} Parts..."
+
+        parts = []
+        File.open(src, "r") do |file|
+          part_number = 0
+          while (data = file.read(chunk_bytes))
+            part_number += 1
+            md5 = Base64.encode64(Digest::MD5.digest(data)).chomp
+
+            with_retries("PUT '#{bucket}/#{dest}' Part ##{part_number}") do
+              resp = connection.upload_part(
+                bucket, dest, upload_id, part_number, StringIO.new(data),
+                "Content-MD5" => md5
+              )
+              parts << resp.headers["ETag"]
+            end
+
+            if (i = progress.rindex(part_number))
+              Logger.info "\s\s...#{i + 1}0% Complete..."
+            end
+          end
+        end
+        parts
+      end
+
+      def complete_multipart(dest, upload_id, parts)
+        Logger.info "\s\sComplete Multipart '#{bucket}/#{dest}'"
+
+        with_retries("POST '#{bucket}/#{dest}' (Complete)") do
+          resp = connection.complete_multipart_upload(bucket, dest,
+                                                      upload_id, parts)
+          raise Error, <<-EOS if resp.body["Code"]
+            The server returned the following error:
+            #{resp.body["Code"]}: #{resp.body["Message"]}
+          EOS
+        end
+      end
+
+      def headers
+        headers = {}
+
+        enc = encryption.to_s.upcase
+        headers["x-amz-server-side-encryption"] = enc unless enc.empty?
+
+        sc = storage_class.to_s.upcase
+        headers["x-amz-storage-class"] = sc unless sc.empty? || sc == "STANDARD"
+
+        headers
+      end
+
+      def adjusted_chunk_bytes(chunk_bytes, file_size)
+        return chunk_bytes if file_size / chunk_bytes.to_f <= 10_000
+
+        mb = orig_mb = chunk_bytes / 1024**2
+        mb += 1 until file_size / (1024**2 * mb).to_f <= 10_000
+        Logger.warn Error.new(<<-EOS)
+          Chunk Size Adjusted
+          Your original #chunk_size of #{orig_mb} MiB has been adjusted
+          to #{mb} MiB in order to satisfy the limit of 10,000 chunks.
+          To enforce your chosen #chunk_size, you should use the Splitter.
+          e.g. split_into_chunks_of #{mb * 10_000} (#chunk_size * 10_000)
+        EOS
+        1024**2 * mb
+      end
+
+      class Object
+        attr_reader :key, :etag, :storage_class
+
+        def initialize(cloud_io, data)
+          @cloud_io = cloud_io
+          @key = data["Key"]
+          @etag = data["ETag"]
+          @storage_class = data["StorageClass"]
+        end
+
+        # currently 'AES256' or nil
+        def encryption
+          metadata["x-amz-server-side-encryption"]
+        end
+
+        private
+
+        def metadata
+          @metadata ||= @cloud_io.head_object(self).headers
+        end
+      end
+    end
+  end
+end

data/lib/backup/compressor/base.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module Backup
+  module Compressor
+    class Base
+      include Utilities::Helpers
+      include Config::Helpers
+
+      ##
+      # Yields to the block the compressor command and filename extension.
+      def compress_with
+        log!
+        yield @cmd, @ext
+      end
+
+      private
+
+      ##
+      # Return the compressor name, with Backup namespace removed
+      def compressor_name
+        self.class.to_s.sub("Backup::", "")
+      end
+
+      ##
+      # Logs a message to the console and log file to inform
+      # the client that Backup is using the compressor
+      def log!
+        Logger.info "Using #{compressor_name} for compression.\n" \
+          "  Command: '#{@cmd}'\n" \
+          "  Ext: '#{@ext}'"
+      end
+    end
+  end
+end

data/lib/backup/compressor/bzip2.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module Backup
+  module Compressor
+    class Bzip2 < Base
+      ##
+      # Specify the level of compression to use.
+      #
+      # Values should be a single digit from 1 to 9.
+      # Note that setting the level to either extreme may or may not
+      # give the desired result. Be sure to check the documentation
+      # for the compressor being used.
+      #
+      # The default `level` is 9.
+      attr_accessor :level
+
+      ##
+      # Creates a new instance of Backup::Compressor::Bzip2
+      def initialize(&block)
+        load_defaults!
+
+        @level ||= false
+
+        instance_eval(&block) if block_given?
+
+        @cmd = "#{utility(:bzip2)}#{options}"
+        @ext = ".bz2"
+      end
+
+      private
+
+      def options
+        " -#{@level}" if @level
+      end
+    end
+  end
+end

data/lib/backup/compressor/custom.rb

@@ -0,0 +1,51 @@
+# frozen_string_literal: true
+
+module Backup
+  module Compressor
+    class Custom < Base
+      ##
+      # Specify the system command to invoke a compressor,
+      # including any command-line arguments.
+      # e.g. @compressor.command = 'pbzip2 -p2 -4'
+      #
+      # The data to be compressed will be piped to the command's STDIN,
+      # and it should write the compressed data to STDOUT.
+      # i.e. `cat file.tar | %command% > file.tar.%extension%`
+      attr_accessor :command
+
+      ##
+      # File extension to append to the compressed file's filename.
+      # e.g. @compressor.extension = '.bz2'
+      attr_accessor :extension
+
+      ##
+      # Initializes a new custom compressor.
+      def initialize(&block)
+        load_defaults!
+
+        instance_eval(&block) if block_given?
+
+        @cmd = set_cmd
+        @ext = set_ext
+      end
+
+      private
+
+      ##
+      # Return the command line using the full path.
+      # Ensures the command exists and is executable.
+      def set_cmd
+        parts = @command.to_s.split(" ")
+        parts[0] = utility(parts[0])
+        parts.join(" ")
+      end
+
+      ##
+      # Return the extension given without whitespace.
+      # If extension was not set, return an empty string
+      def set_ext
+        @extension.to_s.strip
+      end
+    end
+  end
+end