Class: BackupTool
- Inherits:
-
Object
- Object
- BackupTool
- Defined in:
- lib/backuptool.rb
Instance Method Summary collapse
-
#buffered_download(remote, local) ⇒ Object
Download a file from HDFS, buffered way * Args : -
remote
-> HDFS path -local
-> local path. -
#cleanup(days) ⇒ Object
Cleans up backups that are older than a number of days.
-
#create_backup_flag(date) ⇒ Object
Method that creates a backup flag to signal that the backup is finished on all nodes This is an individual command that has to be called manually after snapshots have finished.
- #data_dir_for_node(node) ⇒ Object
- #delete_snapshots(node: @cassandra.node_name, date: 'ALL') ⇒ Object
- #get_backup_flags ⇒ Object
-
#get_node_list(node) ⇒ Object
Get the list of nodes.
-
#get_snapshots_node(node, date) ⇒ Object
Look for all snapshots already existing for “node” at time “date”.
-
#initialize(cassandra, hadoop, logger) ⇒ BackupTool
constructor
Create a new BackupTool instance * Args : -
cassandra
-> Cassandra instance -hadoop
-> HDFS instance -logger
-> Logger. - #list_snapshots(node: @cassandra.node_name) ⇒ Object
- #metadata_dir ⇒ Object
- #metadata_dir_for_backup(node, date) ⇒ Object
- #metadata_dir_for_node(node) ⇒ Object
- #new_snapshot ⇒ Object
- #prepare_hdfs_dirs(node) ⇒ Object
-
#restore_snapshot(node, date, destination, keyspace: 'ALL', table: 'ALL') ⇒ Object
Restore a snapshot from HDFS * Args : -
node
-> node where the snapshot comes from -date
-> snapshot date -destination
-> local directory where to restore. -
#search_snapshots(node: 'ALL', date: 'ALL') ⇒ Object
Look for snapshots * Args : -
node
-> Cassandra node name -date
-> HDFS instance.
Constructor Details
#initialize(cassandra, hadoop, logger) ⇒ BackupTool
Create a new BackupTool instance
-
Args :
-
cassandra
-> Cassandra instance -
hadoop
-> HDFS instance -
logger
-> Logger
-
17 18 19 20 21 22 23 |
# File 'lib/backuptool.rb', line 17 def initialize(cassandra, hadoop, logger) @cassandra = cassandra @hadoop = hadoop @logger = logger @metadir = META_DIR end |
Instance Method Details
#buffered_download(remote, local) ⇒ Object
Download a file from HDFS, buffered way
-
Args :
-
remote
-> HDFS path -
local
-> local path
-
251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 |
# File 'lib/backuptool.rb', line 251 def buffered_download(remote, local) @logger.debug("#{remote} => #{local}") # Create the destination directory if not exists path = File.dirname(local) FileUtils.mkdir_p(path) unless File.exist?(path) file = open(local, 'wb') offset = 0 length = BUFFER_SIZE print '[' while length == BUFFER_SIZE print '#' content = @hadoop.read(remote, offset: offset, length: BUFFER_SIZE) file.write(content) length = content.length offset += length end print "]\n" file.close end |
#cleanup(days) ⇒ Object
Cleans up backups that are older than a number of days. This functions cleans data on all nodes.
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 |
# File 'lib/backuptool.rb', line 204 def cleanup(days) retention_date = Date.today - days @logger.info("Cleaning backup data on all nodes before #{retention_date}.") all_snapshots = search_snapshots @logger.info("A total of #{all_snapshots.size} snapshots were found on HDFS.") snapshots_to_be_deleted = all_snapshots.select { |snapshot| snapshot.get_date < retention_date } @logger.info("A total of #{snapshots_to_be_deleted.size} snapshots will be deleted.") snapshots_to_be_deleted.each do |snapshot| delete_snapshots(node: snapshot.node, date: snapshot.date) end all_backup_flags = get_backup_flags @logger.info("A total of #{all_backup_flags.size} back up flags were found on HDFS.") backup_flags_to_be_delete = all_backup_flags.select { |flag| flag.date < retention_date } @logger.info("A total of #{backup_flags_to_be_delete.size} backup flags will be deleted.") backup_flags_to_be_delete.each do |flag| file = () + flag.file @logger.info("Deleting #{file}") @hadoop.delete(file) end end |
#create_backup_flag(date) ⇒ Object
Method that creates a backup flag to signal that the backup is finished on all nodes This is an individual command that has to be called manually after snapshots have finished
233 234 235 236 237 238 239 |
# File 'lib/backuptool.rb', line 233 def create_backup_flag(date) file_name = 'BACKUP_COMPLETED_' + date remote_file = () + file_name @logger.info('Setting backup completed flag : ' + remote_file) @hadoop.create(remote_file, '', overwrite: true) end |
#data_dir_for_node(node) ⇒ Object
37 38 39 |
# File 'lib/backuptool.rb', line 37 def data_dir_for_node(node) return @hadoop.base_dir + '/' + @cassandra.cluster_name + '/' + node + '/' end |
#delete_snapshots(node: @cassandra.node_name, date: 'ALL') ⇒ Object
174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 |
# File 'lib/backuptool.rb', line 174 def delete_snapshots(node: @cassandra.node_name, date: 'ALL') snapshots = search_snapshots(node: node, date: date) if snapshots.empty? raise('No snapshot found for deletion') end snapshots.each do |snapshot| @logger.info("Deleting snapshot #{snapshot}") node_snapshots = search_snapshots(node: snapshot.node) = Set.new node_snapshots.each do |s| += s. if s != snapshot end files = snapshot. - @logger.info("#{files.length} files to delete") files.each do |file| @logger.info("Deleting file #{file}") remote = data_dir_for_node(snapshot.node) + '/' + file @logger.debug("DELETE => #{remote}") @hadoop.delete(remote) end @logger.info('Deleting metadata in HDFS') remote = (snapshot.node, snapshot.date) @logger.debug("DELETE => #{remote}") @hadoop.delete(remote) end end |
#get_backup_flags ⇒ Object
241 242 243 244 245 |
# File 'lib/backuptool.rb', line 241 def get_backup_flags @hadoop.list(()) .select { |item| item['pathSuffix'].include? 'BACKUP_COMPLETED_' } .collect { |file| BackupFlag.new(@cassandra.cluster_name, file['pathSuffix']) } end |
#get_node_list(node) ⇒ Object
Get the list of nodes
71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 |
# File 'lib/backuptool.rb', line 71 def get_node_list(node) if node != 'ALL' return [node] end nodes = [] begin nodes = @hadoop.list(()) .select { |item| item['type'].casecmp('DIRECTORY') == 0 } .map { |item| item['pathSuffix'] } .flatten rescue Exception => e @logger.warn("Could not get node list for cluster #{@cassandra.cluster_name} : #{e.}") end return nodes end |
#get_snapshots_node(node, date) ⇒ Object
Look for all snapshots already existing for “node” at time “date”
48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 |
# File 'lib/backuptool.rb', line 48 def get_snapshots_node(node, date) results = [] dates = [date] begin if date == 'ALL' dates = @hadoop.list((node)) .select { |dir| dir['pathSuffix'].include? 'cass_snap_' } .map { |dir| dir['pathSuffix'].gsub('cass_snap_', '')} end dates.each do |date| = @hadoop.read((node, date)).split("\n").to_set results.push(CassandraSnapshot.new(@cassandra.cluster_name, node, date, )) end rescue Exception => e @logger.warn("Could not get snapshots for node #{node} : #{e.}") end return results end |
#list_snapshots(node: @cassandra.node_name) ⇒ Object
99 100 101 102 103 |
# File 'lib/backuptool.rb', line 99 def list_snapshots(node: @cassandra.node_name) @logger.info('Listing available snapshots') snapshots = search_snapshots(node: node) tp(snapshots, 'cluster', 'node', 'date') end |
#metadata_dir ⇒ Object
33 34 35 |
# File 'lib/backuptool.rb', line 33 def () return @hadoop.base_dir + '/' + @metadir + '/' + @cassandra.cluster_name + '/' end |
#metadata_dir_for_backup(node, date) ⇒ Object
25 26 27 |
# File 'lib/backuptool.rb', line 25 def (node, date) return () + node + '/cass_snap_' + date + '/' end |
#metadata_dir_for_node(node) ⇒ Object
29 30 31 |
# File 'lib/backuptool.rb', line 29 def (node) return () + node + '/' end |
#new_snapshot ⇒ Object
120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 |
# File 'lib/backuptool.rb', line 120 def new_snapshot @logger.info(':::::::: Creating new snapshot ::::::::') snapshot = @cassandra.new_snapshot prepare_hdfs_dirs(snapshot.node) @logger.info(':::::::: Get last backup ::::::::') existing = search_snapshots(node: snapshot.node) last = if existing.empty? then CassandraSnapshot.new(snapshot.cluster, snapshot.node, 'never') else existing[-1] end @logger.info("Last snapshot is #{last}") files = snapshot. - last. @logger.info("#{files.length} files to upload") @logger.info('::::::: Uploading tables to HDFS ::::::') index = 0 number_of_files = files.length total_file_size = 0 files.each do |file| index += 1 local = @cassandra.data_path + '/' + file local_file_size = File.size(local) total_file_size += local_file_size pretty_size = Filesize.from("#{local_file_size} B").pretty @logger.info("Sending file #{index}/#{number_of_files} #{file} having size #{pretty_size} to HDFS") remote = data_dir_for_node(snapshot.node) + file @logger.debug("#{local} => #{remote}") File.open(local, 'r') do |f| begin retries = 3 @hadoop.create(remote, f, overwrite: true) rescue Exception => e @logger.info("HDFS write failed: #{e.}") @logger.info("HDFS write retrying in 1s") sleep 1 retry if (retries -= 1) < 0 end end end total_file_size_pretty = Filesize.from("#{total_file_size} B").pretty @logger.info("Total size of uploaded files is #{total_file_size_pretty}") @logger.info('Sending metadata to HDFS') remote = (snapshot.node, snapshot.date) @logger.debug("metadata => #{remote}") @hadoop.create(remote, snapshot..to_a * "\n", overwrite: true) @cassandra.delete_snapshot(snapshot) @logger.info('Success !') end |
#prepare_hdfs_dirs(node) ⇒ Object
105 106 107 108 109 110 111 112 113 114 115 116 117 118 |
# File 'lib/backuptool.rb', line 105 def prepare_hdfs_dirs(node) @logger.info(':::::::: Prepare HDFS ::::::::') begin paths = [data_dir_for_node(node), (node)] paths.each do |path| @logger.info("Creating destination directory " + path) if not @hadoop.mkdir(path) raise end end rescue Exception => e raise("Could not create your cluster directory : #{e.}") end end |
#restore_snapshot(node, date, destination, keyspace: 'ALL', table: 'ALL') ⇒ Object
Restore a snapshot from HDFS
-
Args :
-
node
-> node where the snapshot comes from -
date
-> snapshot date -
destination
-> local directory where to restore
-
280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 |
# File 'lib/backuptool.rb', line 280 def restore_snapshot(node, date, destination, keyspace: 'ALL', table: 'ALL') # Search the snapshot matching node and date snapshots = search_snapshots(node: node, date: date) if snapshots.empty? raise('No snapshot found for restore') elsif snapshots.length > 1 raise('More than one candidate snapshot to restore') else snapshot = snapshots[0] @logger.info("Restoring snapshot #{snapshot}") @logger.info("Snapshot has #{snapshot..length} files") files_to_be_restored = snapshot..select { |item| filename = File.basename(item) matches_keyspace = keyspace == 'ALL' || (filename.include? keyspace) matches_table = table == 'ALL' || (filename.include? table) matches_keyspace && matches_table } @logger.info("Found #{files_to_be_restored.length} to be restored that match keyspace #{keyspace} and table #{table}") # For each file in the list files_to_be_restored.each do |file| @logger.info("Restoring file #{file}") local = destination + '/' + file remote = data_dir_for_node(snapshot.node) + file # Download the file from hdfs buffered_download(remote, local) end @logger.info('Success !') end end |
#search_snapshots(node: 'ALL', date: 'ALL') ⇒ Object
Look for snapshots
-
Args :
-
node
-> Cassandra node name -
date
-> HDFS instance
-
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 |
# File 'lib/backuptool.rb', line 45 def search_snapshots(node: 'ALL', date: 'ALL') # Look for all snapshots already existing for "node" at time "date" def get_snapshots_node(node, date) results = [] dates = [date] begin if date == 'ALL' dates = @hadoop.list((node)) .select { |dir| dir['pathSuffix'].include? 'cass_snap_' } .map { |dir| dir['pathSuffix'].gsub('cass_snap_', '')} end dates.each do |date| = @hadoop.read((node, date)).split("\n").to_set results.push(CassandraSnapshot.new(@cassandra.cluster_name, node, date, )) end rescue Exception => e @logger.warn("Could not get snapshots for node #{node} : #{e.}") end return results end # Get the list of nodes def get_node_list(node) if node != 'ALL' return [node] end nodes = [] begin nodes = @hadoop.list(()) .select { |item| item['type'].casecmp('DIRECTORY') == 0 } .map { |item| item['pathSuffix'] } .flatten rescue Exception => e @logger.warn("Could not get node list for cluster #{@cassandra.cluster_name} : #{e.}") end return nodes end # RUN @logger.info("Searching snapshots for #{node} at time #{date}") snapshots = get_node_list(node).map { |node| get_snapshots_node(node, date) } .flatten .sort @logger.info("Found #{snapshots.length} snapshots") return snapshots end |