Merge pull request #73 from 5apps/features/redis_migration

Migration script for copying metadata to Redis

lib/remote_storage/swift.rb

@@ -88,11 +88,11 @@ module RemoteStorage
       lua_script = <<-EOF
         local user = ARGV[1]
         local directory = ARGV[2]
-        local items = redis.call("smembers", "rs_meta:"..user..":"..directory.."/:items")
+        local items = redis.call("smembers", "rs:m:"..user..":"..directory.."/:items")
         local listing = {}
 
         for index, name in pairs(items) do
-          local redis_key = "rs_meta:"..user..":"
+          local redis_key = "rs:m:"..user..":"
           if directory == "" then
             redis_key = redis_key..name
           else
@@ -108,10 +108,10 @@ module RemoteStorage
             metadata[metadata_values[idx]] = metadata_values[idx + 1]
           end
 
-          listing[name] = {["ETag"] = metadata["etag"]}
+          listing[name] = {["ETag"] = metadata["e"]}
           if string.sub(name, -1) ~= "/" then
-            listing[name]["Content-Type"]   = metadata["type"]
+            listing[name]["Content-Type"]   = metadata["t"]
-            listing[name]["Content-Length"] = tonumber(metadata["size"])
+            listing[name]["Content-Length"] = tonumber(metadata["s"])
           end
         end
 
@@ -122,7 +122,7 @@ module RemoteStorage
     end
 
     def get_directory_listing_from_redis(user, directory)
-      etag = redis.hget "rs_meta:#{user}:#{directory}/", "etag"
+      etag = redis.hget "rs:m:#{user}:#{directory}/", "e"
 
       none_match = (server.env["HTTP_IF_NONE_MATCH"] || "").split(",").map(&:strip)
       server.halt 304 if none_match.include? etag
@@ -189,16 +189,18 @@ module RemoteStorage
 
       res = do_put_request(url, data, content_type)
 
-      # TODO get last modified from response and add to metadata
+      # TODO use actual last modified time from the document put request
+      timestamp = (Time.now.to_f * 1000).to_i
+
       metadata = {
-        etag: res.headers[:etag],
+        e: res.headers[:etag],
-        size: data.size,
+        s: data.size,
-        type: content_type
+        t: content_type,
+        m: timestamp
       }
 
       if update_metadata_object(user, directory, key, metadata) &&
-         # TODO provide the last modified to use for the dir objects as well
+         update_dir_objects(user, directory, timestamp)
-         update_dir_objects(user, directory)
         server.headers["ETag"] = %Q("#{res.headers[:etag]}")
         server.halt 200
       else
@@ -312,23 +314,23 @@ module RemoteStorage
         end
 
         -- check for existing directory with the same name as the document
-        local redis_key = "rs_meta:"..user..":"
+        local redis_key = "rs:m:"..user..":"
         if directory == "" then
           redis_key = redis_key..key.."/"
         else
           redis_key = redis_key..directory.."/"..key.."/"
         end
-        if redis.call("hget", redis_key, "etag") then
+        if redis.call("hget", redis_key, "e") then
           return true
         end
 
         for index, dir in pairs(parent_directories) do
-          if redis.call("hget", "rs_meta:"..user..":"..dir.."/", "etag") then
+          if redis.call("hget", "rs:m:"..user..":"..dir.."/", "e") then
             -- the directory already exists, no need to do further checks
             return false
           else
             -- check for existing document with same name as directory
-            if redis.call("hget", "rs_meta:"..user..":"..dir, "etag") then
+            if redis.call("hget", "rs:m:"..user..":"..dir, "e") then
               return true
             end
           end
@@ -400,17 +402,14 @@ module RemoteStorage
     end
 
     def update_metadata_object(user, directory, key, metadata)
-      redis_key = "rs_meta:#{user}:#{directory}/#{key}"
+      redis_key = "rs:m:#{user}:#{directory}/#{key}"
       redis.hmset(redis_key, *metadata)
-      redis.sadd "rs_meta:#{user}:#{directory}/:items", key
+      redis.sadd "rs:m:#{user}:#{directory}/:items", key
 
       true
     end
 
-    def update_dir_objects(user, directory)
+    def update_dir_objects(user, directory, timestamp)
-      # TODO use actual last modified time from the document put request
-      timestamp = (Time.now.to_f * 1000).to_i
-
       parent_directories_for(directory).each do |dir|
         unless dir == ""
           res = do_put_request("#{url_for_directory(user, dir)}/", timestamp.to_s, "text/plain")
@@ -420,10 +419,10 @@ module RemoteStorage
           etag = etag_for(get_response.body)
         end
 
-        key = "rs_meta:#{user}:#{dir}/"
+        key = "rs:m:#{user}:#{dir}/"
-        metadata = {etag: etag, modified: timestamp}
+        metadata = {e: etag, m: timestamp}
         redis.hmset(key, *metadata)
-        redis.sadd "rs_meta:#{user}:#{parent_directory_for(dir)}:items", "#{top_directory(dir)}/"
+        redis.sadd "rs:m:#{user}:#{parent_directory_for(dir)}:items", "#{top_directory(dir)}/"
       end
 
       true
@@ -438,21 +437,22 @@ module RemoteStorage
     end
 
     def delete_metadata_objects(user, directory, key)
-      redis_key = "rs_meta:#{user}:#{directory}/#{key}"
+      redis_key = "rs:m:#{user}:#{directory}/#{key}"
       redis.del(redis_key)
-      redis.srem "rs_meta:#{user}:#{directory}/:items", key
+      redis.srem "rs:m:#{user}:#{directory}/:items", key
     end
 
     def delete_dir_objects(user, directory)
+      timestamp = (Time.now.to_f * 1000).to_i
+
       parent_directories_for(directory).each do |dir|
         if dir_empty?(user, dir)
           unless dir == ""
             do_delete_request("#{url_for_directory(user, dir)}/")
           end
-          redis.del "rs_meta:#{user}:#{directory}/"
+          redis.del "rs:m:#{user}:#{directory}/"
-          redis.srem "rs_meta:#{user}:#{parent_directory_for(dir)}:items", "#{dir}/"
+          redis.srem "rs:m:#{user}:#{parent_directory_for(dir)}:items", "#{dir}/"
         else
-          timestamp = (Time.now.to_f * 1000).to_i
           unless dir == ""
             res = do_put_request("#{url_for_directory(user, dir)}/", timestamp.to_s, "text/plain")
             etag = res.headers[:etag]
@@ -460,15 +460,15 @@ module RemoteStorage
             get_response = do_get_request("#{container_url_for(user)}/?format=json&path=")
             etag = etag_for(get_response.body)
           end
-          metadata = {etag: etag, modified: timestamp}
+          metadata = {e: etag, m: timestamp}
-          redis.hmset("rs_meta:#{user}:#{dir}/", *metadata)
+          redis.hmset("rs:m:#{user}:#{dir}/", *metadata)
         end
       end
     end
 
     def dir_empty?(user, dir)
       if directory_backend(user).match(/new/)
-        redis.smembers("rs_meta:#{user}:#{dir}/:items").empty?
+        redis.smembers("rs:m:#{user}:#{dir}/:items").empty?
       else
         do_get_request("#{container_url_for(user)}/?format=plain&limit=1&path=#{escape(dir)}/") do |res|
           return res.headers[:content_length] == "0"
@@ -538,7 +538,7 @@ module RemoteStorage
     end
 
     def directory_backend(user)
-      @directory_backend ||= redis.get("rs_config:dir_backend:#{user}") || "legacy"
+      @directory_backend ||= redis.get("rsc:db:#{user}") || "legacy"
     end
 
     def etag_for(body)

migrate_metadata_to_redis.rb

@@ -0,0 +1,246 @@
+#!/usr/bin/env ruby
+
+require "rest_client"
+require "redis"
+require "yaml"
+require "logger"
+require "active_support/core_ext/hash"
+
+class Migrator
+
+  attr_accessor :username, :base_url, :swift_host, :swift_token,
+                :environment, :dry_run, :settings, :logger
+
+  def initialize(username)
+    @username = username
+
+    @environment = ENV["ENVIRONMENT"] || "staging"
+    @settings = YAML.load(File.read('config.yml'))[@environment]
+
+    @swift_host = @settings["swift"]["host"]
+    @swift_token = File.read("tmp/swift_token.txt").strip
+
+    @dry_run = ENV["DRYRUN"] || false # disables writing anything to Redis when true
+
+    @logger = Logger.new("log/migrate_metadata_to_redis.log")
+    log_level = ENV["LOGLEVEL"] || "INFO"
+    logger.level = Kernel.const_get "Logger::#{log_level}"
+    logger.progname = username
+  end
+
+  def root_url
+    "#{@base_url}/#{@username}"
+  end
+
+  def is_dir?(name)
+    name[-1] == "/"
+  end
+
+  def url_for(directory, parent_directory="")
+    "#{root_url}#{parent_directory}#{directory}"
+  end
+
+  def migrate
+    logger.info "Starting migration for '#{username}'"
+    set_directory_backend("legacy_locked")
+    begin
+      work_on_dir("", "")
+    rescue Exception => ex
+      logger.error "Error migrating metadata for '#{username}': #{ex}"
+      set_directory_backend("legacy")
+      # write username to file for later reference
+      File.open('log/failed_migration.log', 'a') { |f| f.puts username }
+      exit 1
+    end
+    set_directory_backend("new")
+    logger.info "Finished migration for '#{username}'"
+  end
+
+  def set_directory_backend(backend)
+    redis.set("rsc:db:#{username}", backend) unless dry_run
+  end
+
+  def work_on_dir(directory, parent_directory)
+    logger.debug "Retrieving listing for '#{parent_directory}#{directory}'"
+
+    listing = get_directory_listing_from_swift("#{parent_directory}#{directory}")
+
+    timestamp = (Time.now.to_f * 1000).to_i
+
+    if listing["items"].any?
+      items = listing["items"]
+      items.each do |item, data|
+        if is_dir? item
+          save_directory_data("#{parent_directory}#{directory}", item, data, timestamp)
+
+          # get dir listing and repeat
+          work_on_dir(item, "#{parent_directory}#{directory}")
+        else
+          save_document_data("#{parent_directory}#{directory}", item, data)
+        end
+
+        add_item_to_parent_dir("#{parent_directory}#{directory}", item)
+      end
+    end
+  end
+
+  def add_item_to_parent_dir(dir, item)
+    key = "rs:m:#{username}:#{parent_directory_for(dir)}:items"
+    logger.debug "Adding item #{item} to #{key}"
+    redis.sadd(key, item) unless dry_run
+  end
+
+  def save_directory_data(dir, item, data, timestamp)
+    key = "rs:m:#{username}:#{dir.gsub(/^\//, "")}#{item}"
+    metadata = {
+      e: data["ETag"],
+      m: timestamp_for(data["Last-Modified"])
+    }
+
+    logger.debug "Metadata for dir #{key}: #{metadata}"
+    redis.hmset(key, *metadata) unless dry_run
+  end
+
+  def save_document_data(dir, item, data)
+    key = "rs:m:#{username}:#{dir.gsub(/^\//, "")}#{item}"
+    metadata = {
+      e: data["ETag"],
+      s: data["Content-Length"],
+      t: data["Content-Type"],
+      m: timestamp_for(data["Last-Modified"])
+    }
+    logger.debug "Metadata for document #{key}: #{metadata}"
+    redis.hmset(key, *metadata) unless dry_run
+  end
+
+  def parent_directory_for(directory)
+    if directory.match(/\//)
+      return directory[0..directory.rindex("/")]
+    else
+      return "/"
+    end
+  end
+
+  def timestamp_for(date)
+    return DateTime.parse(date).strftime("%Q").to_i
+  end
+
+  def redis
+    @redis ||= Redis.new(@settings["redis"].symbolize_keys)
+  end
+
+  def get_directory_listing_from_swift(directory)
+    is_root_listing = directory.empty?
+
+    get_response = nil
+
+    do_head_request("#{url_for_directory(@username, directory)}") do |response|
+      return directory_listing([]) if response.code == 404
+
+      if is_root_listing
+        get_response = do_get_request("#{container_url_for(@username)}/?format=json&path=")
+      else
+        get_response = do_get_request("#{container_url_for(@username)}/?format=json&path=#{escape(directory)}")
+      end
+    end
+
+    if body = JSON.parse(get_response.body)
+      listing = directory_listing(body)
+    else
+      puts "listing not JSON"
+    end
+
+    listing
+  end
+
+  def directory_listing(res_body)
+    listing = {
+      "@context" => "http://remotestorage.io/spec/folder-description",
+      "items"    => {}
+    }
+
+    res_body.each do |entry|
+      name = entry["name"]
+      name.sub!("#{File.dirname(entry["name"])}/", '')
+      if name[-1] == "/" # It's a directory
+        listing["items"].merge!({
+          name => {
+            "ETag"           => entry["hash"],
+            "Last-Modified"  => entry["last_modified"]
+          }
+        })
+      else # It's a file
+        listing["items"].merge!({
+          name => {
+            "ETag"           => entry["hash"],
+            "Content-Type"   => entry["content_type"],
+            "Content-Length" => entry["bytes"],
+            "Last-Modified"  => entry["last_modified"]
+          }
+        })
+      end
+    end
+
+    listing
+  end
+
+  def etag_for(body)
+    objects = JSON.parse(body)
+
+    if objects.empty?
+      Digest::MD5.hexdigest ""
+    else
+      Digest::MD5.hexdigest objects.map { |o| o["hash"] }.join
+    end
+  end
+
+  def do_head_request(url, &block)
+    RestClient.head(url, default_headers, &block)
+  end
+
+  def do_get_request(url, &block)
+    RestClient.get(url, default_headers, &block)
+  end
+
+  def default_headers
+    {"x-auth-token" => @swift_token}
+  end
+
+  def url_for_directory(user, directory)
+    if directory.empty?
+      container_url_for(user)
+    else
+      "#{container_url_for(user)}/#{escape(directory)}"
+    end
+  end
+
+  def container_url_for(user)
+    "#{base_url}/#{container_for(user)}"
+  end
+
+  def base_url
+    @base_url ||= @swift_host
+  end
+
+  def container_for(user)
+    "rs:#{environment.to_s.chars.first}:#{user}"
+  end
+
+  def escape(url)
+    # We want spaces to turn into %20 and slashes to stay slashes
+    CGI::escape(url).gsub('+', '%20').gsub('%2F', '/')
+  end
+end
+
+username = ARGV[0]
+
+unless username
+  puts "No username given."
+  puts "Usage:"
+  puts "ENVIRONMENT=staging ./migrate_metadata_to_redis.rb <username>"
+  exit 1
+end
+
+migrator = Migrator.new username
+migrator.migrate
+

spec/spec_helper.rb

@@ -39,7 +39,7 @@ if app.settings.respond_to? :redis
   end
 
   def purge_redis
-    redis.keys("rs_*").each do |key|
+    redis.keys("rs*").each do |key|
       redis.del key
     end
   end

spec/swift/app_spec.rb

@@ -16,7 +16,7 @@ describe "App" do
 
     before do
       purge_redis
-      redis.set "rs_config:dir_backend:phil", "new"
+      redis.set "rsc:db:phil", "new"
     end
 
     context "authorized" do
@@ -31,11 +31,11 @@ describe "App" do
           put "/phil/food/aguacate", "si"
         end
 
-        metadata = redis.hgetall "rs_meta:phil:food/aguacate"
+        metadata = redis.hgetall "rs:m:phil:food/aguacate"
-        metadata["size"].must_equal "2"
+        metadata["s"].must_equal "2"
-        metadata["type"].must_equal "text/plain; charset=utf-8"
+        metadata["t"].must_equal "text/plain; charset=utf-8"
-        metadata["etag"].must_equal "bla"
+        metadata["e"].must_equal "bla"
-        metadata["modified"].must_equal nil
+        metadata["m"].length.must_equal 13
       end
 
       it "creates the directory objects metadata in redis" do
@@ -50,20 +50,20 @@ describe "App" do
           end
         end
 
-        metadata = redis.hgetall "rs_meta:phil:/"
+        metadata = redis.hgetall "rs:m:phil:/"
-        metadata["etag"].must_equal "rootetag"
+        metadata["e"].must_equal "rootetag"
-        metadata["modified"].length.must_equal 13
+        metadata["m"].length.must_equal 13
 
-        metadata = redis.hgetall "rs_meta:phil:food/"
+        metadata = redis.hgetall "rs:m:phil:food/"
-        metadata["etag"].must_equal "bla"
+        metadata["e"].must_equal "bla"
-        metadata["modified"].length.must_equal 13
+        metadata["m"].length.must_equal 13
 
-        food_items = redis.smembers "rs_meta:phil:food/:items"
+        food_items = redis.smembers "rs:m:phil:food/:items"
         food_items.each do |food_item|
           ["camaron", "aguacate"].must_include food_item
         end
 
-        root_items = redis.smembers "rs_meta:phil:/:items"
+        root_items = redis.smembers "rs:m:phil:/:items"
         root_items.must_equal ["food/"]
       end
 
@@ -81,8 +81,8 @@ describe "App" do
 
           last_response.status.must_equal 200
 
-          metadata = redis.hgetall "rs_meta:phil:food/aguacate"
+          metadata = redis.hgetall "rs:m:phil:food/aguacate"
-          metadata["size"].must_equal "2"
+          metadata["s"].must_equal "2"
         end
 
         it "conflicts when there is a directory with same name as document" do
@@ -94,7 +94,7 @@ describe "App" do
 
           last_response.status.must_equal 409
 
-          metadata = redis.hgetall "rs_meta:phil:food"
+          metadata = redis.hgetall "rs:m:phil:food"
           metadata.must_be_empty
         end
 
@@ -107,7 +107,7 @@ describe "App" do
 
           last_response.status.must_equal 409
 
-          metadata = redis.hgetall "rs_meta:phil:food/aguacate/empanado"
+          metadata = redis.hgetall "rs:m:phil:food/aguacate/empanado"
           metadata.must_be_empty
         end
       end
@@ -115,7 +115,7 @@ describe "App" do
       describe "directory backend configuration" do
         context "locked new backed" do
           before do
-            redis.set "rs_config:dir_backend:phil", "new-locked"
+            redis.set "rsc:db:phil", "new-locked"
           end
 
           it "responds with 503" do
@@ -123,14 +123,14 @@ describe "App" do
 
             last_response.status.must_equal 503
 
-            metadata = redis.hgetall "rs_meta:phil:food/aguacate"
+            metadata = redis.hgetall "rs:m:phil:food/aguacate"
             metadata.must_be_empty
           end
         end
 
         context "locked legacy backend" do
           before do
-            redis.set "rs_config:dir_backend:phil", "legacy-locked"
+            redis.set "rsc:db:phil", "legacy-locked"
           end
 
           it "responds with 503" do
@@ -138,7 +138,7 @@ describe "App" do
 
             last_response.status.must_equal 503
 
-            metadata = redis.hgetall "rs_meta:phil:food/aguacate"
+            metadata = redis.hgetall "rs:m:phil:food/aguacate"
             metadata.must_be_empty
           end
         end
@@ -150,7 +150,7 @@ describe "App" do
 
     before do
       purge_redis
-      redis.set "rs_config:dir_backend:phil", "new"
+      redis.set "rsc:db:phil", "new"
     end
 
     context "authorized" do
@@ -178,12 +178,12 @@ describe "App" do
           end
         end
 
-        metadata = redis.hgetall "rs_meta:phil:food/aguacate"
+        metadata = redis.hgetall "rs:m:phil:food/aguacate"
         metadata.must_be_empty
       end
 
       it "deletes the directory objects metadata in redis" do
-        old_metadata = redis.hgetall "rs_meta:phil:food/"
+        old_metadata = redis.hgetall "rs:m:phil:food/"
 
         put_stub = OpenStruct.new(headers: {etag: "newetag"})
         get_stub = OpenStruct.new(body: "rootbody")
@@ -197,15 +197,15 @@ describe "App" do
           end
         end
 
-        metadata = redis.hgetall "rs_meta:phil:food/"
+        metadata = redis.hgetall "rs:m:phil:food/"
-        metadata["etag"].must_equal "newetag"
+        metadata["e"].must_equal "newetag"
-        metadata["modified"].length.must_equal 13
+        metadata["m"].length.must_equal 13
-        metadata["modified"].wont_equal old_metadata["modified"]
+        metadata["m"].wont_equal old_metadata["m"]
 
-        food_items = redis.smembers "rs_meta:phil:food/:items"
+        food_items = redis.smembers "rs:m:phil:food/:items"
         food_items.must_equal ["camaron"]
 
-        root_items = redis.smembers "rs_meta:phil:/:items"
+        root_items = redis.smembers "rs:m:phil:/:items"
         root_items.must_equal ["food/"]
       end
 
@@ -223,13 +223,13 @@ describe "App" do
           end
         end
 
-        metadata = redis.hgetall "rs_meta:phil:food/"
+        metadata = redis.hgetall "rs:m:phil:food/"
         metadata.must_be_empty
 
-        food_items = redis.smembers "rs_meta:phil:food/:items"
+        food_items = redis.smembers "rs:m:phil:food/:items"
         food_items.must_be_empty
 
-        root_items = redis.smembers "rs_meta:phil:/:items"
+        root_items = redis.smembers "rs:m:phil:/:items"
         root_items.must_be_empty
       end
     end
@@ -239,7 +239,7 @@ describe "App" do
 
     before do
       purge_redis
-      redis.set "rs_config:dir_backend:phil", "new"
+      redis.set "rsc:db:phil", "new"
     end
 
     context "authorized" do
@@ -318,7 +318,7 @@ describe "App" do
           put "/phil/food/camaron", "yummi"
         end
 
-        redis.set "rs_config:dir_backend:phil", "legacy"
+        redis.set "rsc:db:phil", "legacy"
       end
 
       it "serves directory listing from Swift backend" do