Pankaj Berde | f5ca08c | 2013-06-10 21:36:34 -0700 | [diff] [blame] | 1 | # Cassandra storage config YAML |
| 2 | |
| 3 | # NOTE: |
| 4 | # See http://wiki.apache.org/cassandra/StorageConfiguration for |
| 5 | # full explanations of configuration directives |
| 6 | # /NOTE |
| 7 | |
| 8 | # The name of the cluster. This is mainly used to prevent machines in |
| 9 | # one logical cluster from joining another. |
| 10 | cluster_name: 'Test Cluster' |
| 11 | |
| 12 | # This defines the number of tokens randomly assigned to this node on the ring |
| 13 | # The more tokens, relative to other nodes, the larger the proportion of data |
| 14 | # that this node will store. You probably want all nodes to have the same number |
| 15 | # of tokens assuming they have equal hardware capability. |
| 16 | # |
| 17 | # If you leave this unspecified, Cassandra will use the default of 1 token for legacy compatibility, |
| 18 | # and will use the initial_token as described below. |
| 19 | # |
| 20 | # Specifying initial_token will override this setting. |
| 21 | # |
| 22 | # If you already have a cluster with 1 token per node, and wish to migrate to |
| 23 | # multiple tokens per node, see http://wiki.apache.org/cassandra/Operations |
| 24 | # num_tokens: 256 |
| 25 | |
| 26 | # If you haven't specified num_tokens, or have set it to the default of 1 then |
| 27 | # you should always specify InitialToken when setting up a production |
| 28 | # cluster for the first time, and often when adding capacity later. |
| 29 | # The principle is that each node should be given an equal slice of |
| 30 | # the token ring; see http://wiki.apache.org/cassandra/Operations |
| 31 | # for more details. |
| 32 | # |
| 33 | # If blank, Cassandra will request a token bisecting the range of |
| 34 | # the heaviest-loaded existing node. If there is no load information |
| 35 | # available, such as is the case with a new cluster, it will pick |
| 36 | # a random token, which will lead to hot spots. |
| 37 | initial_token: |
| 38 | |
| 39 | # See http://wiki.apache.org/cassandra/HintedHandoff |
| 40 | hinted_handoff_enabled: true |
| 41 | # this defines the maximum amount of time a dead host will have hints |
| 42 | # generated. After it has been dead this long, hints will be dropped. |
| 43 | max_hint_window_in_ms: 10800000 # 3 hours |
| 44 | # throttle in KB's per second, per delivery thread |
| 45 | hinted_handoff_throttle_in_kb: 1024 |
| 46 | # Number of threads with which to deliver hints; |
| 47 | # Consider increasing this number when you have multi-dc deployments, since |
| 48 | # cross-dc handoff tends to be slower |
| 49 | max_hints_delivery_threads: 2 |
| 50 | |
| 51 | # The following setting populates the page cache on memtable flush and compaction |
| 52 | # WARNING: Enable this setting only when the whole node's data fits in memory. |
| 53 | # Defaults to: false |
| 54 | # populate_io_cache_on_flush: false |
| 55 | |
| 56 | # authentication backend, implementing IAuthenticator; used to identify users |
| 57 | authenticator: org.apache.cassandra.auth.AllowAllAuthenticator |
| 58 | |
| 59 | # authorization backend, implementing IAuthorizer; used to limit access/provide permissions |
| 60 | authorizer: org.apache.cassandra.auth.AllowAllAuthorizer |
| 61 | |
| 62 | # The partitioner is responsible for distributing rows (by key) across |
| 63 | # nodes in the cluster. Any IPartitioner may be used, including your |
| 64 | # own as long as it is on the classpath. Out of the box, Cassandra |
| 65 | # provides org.apache.cassandra.dht.{Murmur3Partitioner, RandomPartitioner |
| 66 | # ByteOrderedPartitioner, OrderPreservingPartitioner (deprecated)}. |
| 67 | # |
| 68 | # - RandomPartitioner distributes rows across the cluster evenly by md5. |
| 69 | # This is the default prior to 1.2 and is retained for compatibility. |
| 70 | # - Murmur3Partitioner is similar to RandomPartioner but uses Murmur3_128 |
| 71 | # Hash Function instead of md5. When in doubt, this is the best option. |
| 72 | # - ByteOrderedPartitioner orders rows lexically by key bytes. BOP allows |
| 73 | # scanning rows in key order, but the ordering can generate hot spots |
| 74 | # for sequential insertion workloads. |
| 75 | # - OrderPreservingPartitioner is an obsolete form of BOP, that stores |
| 76 | # - keys in a less-efficient format and only works with keys that are |
| 77 | # UTF8-encoded Strings. |
| 78 | # - CollatingOPP colates according to EN,US rules rather than lexical byte |
| 79 | # ordering. Use this as an example if you need custom collation. |
| 80 | # |
| 81 | # See http://wiki.apache.org/cassandra/Operations for more on |
| 82 | # partitioners and token selection. |
| 83 | partitioner: org.apache.cassandra.dht.RandomPartitioner |
| 84 | |
| 85 | # directories where Cassandra should store data on disk. |
| 86 | data_file_directories: |
| 87 | - /tmp/cassandra/data |
| 88 | |
| 89 | # commit log |
| 90 | commitlog_directory: /tmp/cassandra/commitlog |
| 91 | |
| 92 | # policy for data disk failures: |
| 93 | # stop: shut down gossip and Thrift, leaving the node effectively dead, but |
| 94 | # still inspectable via JMX. |
| 95 | # best_effort: stop using the failed disk and respond to requests based on |
| 96 | # remaining available sstables. This means you WILL see obsolete |
| 97 | # data at CL.ONE! |
| 98 | # ignore: ignore fatal errors and let requests fail, as in pre-1.2 Cassandra |
| 99 | disk_failure_policy: stop |
| 100 | |
| 101 | # Maximum size of the key cache in memory. |
| 102 | # |
| 103 | # Each key cache hit saves 1 seek and each row cache hit saves 2 seeks at the |
| 104 | # minimum, sometimes more. The key cache is fairly tiny for the amount of |
| 105 | # time it saves, so it's worthwhile to use it at large numbers. |
| 106 | # The row cache saves even more time, but must store the whole values of |
| 107 | # its rows, so it is extremely space-intensive. It's best to only use the |
| 108 | # row cache if you have hot rows or static rows. |
| 109 | # |
| 110 | # NOTE: if you reduce the size, you may not get you hottest keys loaded on startup. |
| 111 | # |
| 112 | # Default value is empty to make it "auto" (min(5% of Heap (in MB), 100MB)). Set to 0 to disable key cache. |
| 113 | key_cache_size_in_mb: |
| 114 | |
| 115 | # Duration in seconds after which Cassandra should |
| 116 | # safe the keys cache. Caches are saved to saved_caches_directory as |
| 117 | # specified in this configuration file. |
| 118 | # |
| 119 | # Saved caches greatly improve cold-start speeds, and is relatively cheap in |
| 120 | # terms of I/O for the key cache. Row cache saving is much more expensive and |
| 121 | # has limited use. |
| 122 | # |
| 123 | # Default is 14400 or 4 hours. |
| 124 | key_cache_save_period: 14400 |
| 125 | |
| 126 | # Number of keys from the key cache to save |
| 127 | # Disabled by default, meaning all keys are going to be saved |
| 128 | # key_cache_keys_to_save: 100 |
| 129 | |
| 130 | # Maximum size of the row cache in memory. |
| 131 | # NOTE: if you reduce the size, you may not get you hottest keys loaded on startup. |
| 132 | # |
| 133 | # Default value is 0, to disable row caching. |
| 134 | row_cache_size_in_mb: 0 |
| 135 | |
| 136 | # Duration in seconds after which Cassandra should |
| 137 | # safe the row cache. Caches are saved to saved_caches_directory as specified |
| 138 | # in this configuration file. |
| 139 | # |
| 140 | # Saved caches greatly improve cold-start speeds, and is relatively cheap in |
| 141 | # terms of I/O for the key cache. Row cache saving is much more expensive and |
| 142 | # has limited use. |
| 143 | # |
| 144 | # Default is 0 to disable saving the row cache. |
| 145 | row_cache_save_period: 0 |
| 146 | |
| 147 | # Number of keys from the row cache to save |
| 148 | # Disabled by default, meaning all keys are going to be saved |
| 149 | # row_cache_keys_to_save: 100 |
| 150 | |
| 151 | # The provider for the row cache to use. |
| 152 | # |
| 153 | # Supported values are: ConcurrentLinkedHashCacheProvider, SerializingCacheProvider |
| 154 | # |
| 155 | # SerializingCacheProvider serialises the contents of the row and stores |
| 156 | # it in native memory, i.e., off the JVM Heap. Serialized rows take |
| 157 | # significantly less memory than "live" rows in the JVM, so you can cache |
| 158 | # more rows in a given memory footprint. And storing the cache off-heap |
| 159 | # means you can use smaller heap sizes, reducing the impact of GC pauses. |
| 160 | # |
| 161 | # It is also valid to specify the fully-qualified class name to a class |
| 162 | # that implements org.apache.cassandra.cache.IRowCacheProvider. |
| 163 | # |
| 164 | # Defaults to SerializingCacheProvider |
| 165 | row_cache_provider: SerializingCacheProvider |
| 166 | |
| 167 | # saved caches |
| 168 | saved_caches_directory: /tmp/cassandra/saved_caches |
| 169 | |
| 170 | # commitlog_sync may be either "periodic" or "batch." |
| 171 | # When in batch mode, Cassandra won't ack writes until the commit log |
| 172 | # has been fsynced to disk. It will wait up to |
| 173 | # commitlog_sync_batch_window_in_ms milliseconds for other writes, before |
| 174 | # performing the sync. |
| 175 | # |
| 176 | # commitlog_sync: batch |
| 177 | # commitlog_sync_batch_window_in_ms: 50 |
| 178 | # |
| 179 | # the other option is "periodic" where writes may be acked immediately |
| 180 | # and the CommitLog is simply synced every commitlog_sync_period_in_ms |
| 181 | # milliseconds. |
| 182 | commitlog_sync: periodic |
| 183 | commitlog_sync_period_in_ms: 10000 |
| 184 | |
| 185 | # The size of the individual commitlog file segments. A commitlog |
| 186 | # segment may be archived, deleted, or recycled once all the data |
| 187 | # in it (potentally from each columnfamily in the system) has been |
| 188 | # flushed to sstables. |
| 189 | # |
| 190 | # The default size is 32, which is almost always fine, but if you are |
| 191 | # archiving commitlog segments (see commitlog_archiving.properties), |
| 192 | # then you probably want a finer granularity of archiving; 8 or 16 MB |
| 193 | # is reasonable. |
| 194 | commitlog_segment_size_in_mb: 32 |
| 195 | |
| 196 | # any class that implements the SeedProvider interface and has a |
| 197 | # constructor that takes a Map<String, String> of parameters will do. |
| 198 | seed_provider: |
| 199 | # Addresses of hosts that are deemed contact points. |
| 200 | # Cassandra nodes use this list of hosts to find each other and learn |
| 201 | # the topology of the ring. You must change this if you are running |
| 202 | # multiple nodes! |
| 203 | - class_name: org.apache.cassandra.locator.SimpleSeedProvider |
| 204 | parameters: |
| 205 | # seeds is actually a comma-delimited list of addresses. |
| 206 | # Ex: "<ip1>,<ip2>,<ip3>" |
| 207 | - seeds: "127.0.0.1" |
| 208 | |
| 209 | # emergency pressure valve: each time heap usage after a full (CMS) |
| 210 | # garbage collection is above this fraction of the max, Cassandra will |
| 211 | # flush the largest memtables. |
| 212 | # |
| 213 | # Set to 1.0 to disable. Setting this lower than |
| 214 | # CMSInitiatingOccupancyFraction is not likely to be useful. |
| 215 | # |
| 216 | # RELYING ON THIS AS YOUR PRIMARY TUNING MECHANISM WILL WORK POORLY: |
| 217 | # it is most effective under light to moderate load, or read-heavy |
| 218 | # workloads; under truly massive write load, it will often be too |
| 219 | # little, too late. |
| 220 | flush_largest_memtables_at: 0.75 |
| 221 | |
| 222 | # emergency pressure valve #2: the first time heap usage after a full |
| 223 | # (CMS) garbage collection is above this fraction of the max, |
| 224 | # Cassandra will reduce cache maximum _capacity_ to the given fraction |
| 225 | # of the current _size_. Should usually be set substantially above |
| 226 | # flush_largest_memtables_at, since that will have less long-term |
| 227 | # impact on the system. |
| 228 | # |
| 229 | # Set to 1.0 to disable. Setting this lower than |
| 230 | # CMSInitiatingOccupancyFraction is not likely to be useful. |
| 231 | reduce_cache_sizes_at: 0.85 |
| 232 | reduce_cache_capacity_to: 0.6 |
| 233 | |
| 234 | # For workloads with more data than can fit in memory, Cassandra's |
| 235 | # bottleneck will be reads that need to fetch data from |
| 236 | # disk. "concurrent_reads" should be set to (16 * number_of_drives) in |
| 237 | # order to allow the operations to enqueue low enough in the stack |
| 238 | # that the OS and drives can reorder them. |
| 239 | # |
| 240 | # On the other hand, since writes are almost never IO bound, the ideal |
| 241 | # number of "concurrent_writes" is dependent on the number of cores in |
| 242 | # your system; (8 * number_of_cores) is a good rule of thumb. |
| 243 | concurrent_reads: 32 |
| 244 | concurrent_writes: 32 |
| 245 | |
| 246 | # Total memory to use for memtables. Cassandra will flush the largest |
| 247 | # memtable when this much memory is used. |
| 248 | # If omitted, Cassandra will set it to 1/3 of the heap. |
| 249 | # memtable_total_space_in_mb: 2048 |
| 250 | |
| 251 | # Total space to use for commitlogs. Since commitlog segments are |
| 252 | # mmapped, and hence use up address space, the default size is 32 |
| 253 | # on 32-bit JVMs, and 1024 on 64-bit JVMs. |
| 254 | # |
| 255 | # If space gets above this value (it will round up to the next nearest |
| 256 | # segment multiple), Cassandra will flush every dirty CF in the oldest |
| 257 | # segment and remove it. So a small total commitlog space will tend |
| 258 | # to cause more flush activity on less-active columnfamilies. |
| 259 | # commitlog_total_space_in_mb: 4096 |
| 260 | |
| 261 | # This sets the amount of memtable flush writer threads. These will |
| 262 | # be blocked by disk io, and each one will hold a memtable in memory |
| 263 | # while blocked. If you have a large heap and many data directories, |
| 264 | # you can increase this value for better flush performance. |
| 265 | # By default this will be set to the amount of data directories defined. |
| 266 | #memtable_flush_writers: 1 |
| 267 | |
| 268 | # the number of full memtables to allow pending flush, that is, |
| 269 | # waiting for a writer thread. At a minimum, this should be set to |
| 270 | # the maximum number of secondary indexes created on a single CF. |
| 271 | memtable_flush_queue_size: 4 |
| 272 | |
| 273 | # Whether to, when doing sequential writing, fsync() at intervals in |
| 274 | # order to force the operating system to flush the dirty |
| 275 | # buffers. Enable this to avoid sudden dirty buffer flushing from |
| 276 | # impacting read latencies. Almost always a good idea on SSD:s; not |
| 277 | # necessarily on platters. |
| 278 | trickle_fsync: false |
| 279 | trickle_fsync_interval_in_kb: 10240 |
| 280 | |
| 281 | # TCP port, for commands and data |
| 282 | storage_port: 7000 |
| 283 | |
| 284 | # SSL port, for encrypted communication. Unused unless enabled in |
| 285 | # encryption_options |
| 286 | ssl_storage_port: 7001 |
| 287 | |
| 288 | # Address to bind to and tell other Cassandra nodes to connect to. You |
| 289 | # _must_ change this if you want multiple nodes to be able to |
| 290 | # communicate! |
| 291 | # |
| 292 | # Leaving it blank leaves it up to InetAddress.getLocalHost(). This |
| 293 | # will always do the Right Thing *if* the node is properly configured |
| 294 | # (hostname, name resolution, etc), and the Right Thing is to use the |
| 295 | # address associated with the hostname (it might not be). |
| 296 | # |
| 297 | # Setting this to 0.0.0.0 is always wrong. |
| 298 | listen_address: localhost |
| 299 | |
| 300 | # Address to broadcast to other Cassandra nodes |
| 301 | # Leaving this blank will set it to the same value as listen_address |
| 302 | # broadcast_address: 1.2.3.4 |
| 303 | |
| 304 | |
| 305 | # Whether to start the native transport server. |
| 306 | # Currently, only the thrift server is started by default because the native |
| 307 | # transport is considered beta. |
| 308 | # Please note that the address on which the native transport is bound is the |
| 309 | # same as the rpc_address. The port however is different and specified below. |
| 310 | start_native_transport: false |
| 311 | # port for the CQL native transport to listen for clients on |
| 312 | native_transport_port: 9042 |
| 313 | # The minimum and maximum threads for handling requests when the native |
| 314 | # transport is used. The meaning is those is similar to the one of |
| 315 | # rpc_min_threads and rpc_max_threads, though the default differ slightly and |
| 316 | # are the ones below: |
| 317 | # native_transport_min_threads: 16 |
| 318 | # native_transport_max_threads: 128 |
| 319 | |
| 320 | |
| 321 | # Whether to start the thrift rpc server. |
| 322 | start_rpc: true |
| 323 | # The address to bind the Thrift RPC service to -- clients connect |
| 324 | # here. Unlike ListenAddress above, you *can* specify 0.0.0.0 here if |
| 325 | # you want Thrift to listen on all interfaces. |
| 326 | # |
| 327 | # Leaving this blank has the same effect it does for ListenAddress, |
| 328 | # (i.e. it will be based on the configured hostname of the node). |
| 329 | rpc_address: localhost |
| 330 | # port for Thrift to listen for clients on |
| 331 | rpc_port: 9160 |
| 332 | |
| 333 | # enable or disable keepalive on rpc connections |
| 334 | rpc_keepalive: true |
| 335 | |
| 336 | # Cassandra provides three out-of-the-box options for the RPC Server: |
| 337 | # |
| 338 | # sync -> One thread per thrift connection. For a very large number of clients, memory |
| 339 | # will be your limiting factor. On a 64 bit JVM, 128KB is the minimum stack size |
| 340 | # per thread, and that will correspond to your use of virtual memory (but physical memory |
| 341 | # may be limited depending on use of stack space). |
| 342 | # |
| 343 | # hsha -> Stands for "half synchronous, half asynchronous." All thrift clients are handled |
| 344 | # asynchronously using a small number of threads that does not vary with the amount |
| 345 | # of thrift clients (and thus scales well to many clients). The rpc requests are still |
| 346 | # synchronous (one thread per active request). |
| 347 | # |
| 348 | # The default is sync because on Windows hsha is about 30% slower. On Linux, |
| 349 | # sync/hsha performance is about the same, with hsha of course using less memory. |
| 350 | # |
| 351 | # Alternatively, can provide your own RPC server by providing the fully-qualified class name |
| 352 | # of an o.a.c.t.TServerFactory that can create an instance of it. |
| 353 | rpc_server_type: sync |
| 354 | |
| 355 | # Uncomment rpc_min|max_thread to set request pool size limits. |
| 356 | # |
| 357 | # Regardless of your choice of RPC server (see above), the number of maximum requests in the |
| 358 | # RPC thread pool dictates how many concurrent requests are possible (but if you are using the sync |
| 359 | # RPC server, it also dictates the number of clients that can be connected at all). |
| 360 | # |
| 361 | # The default is unlimited and thus provide no protection against clients overwhelming the server. You are |
| 362 | # encouraged to set a maximum that makes sense for you in production, but do keep in mind that |
| 363 | # rpc_max_threads represents the maximum number of client requests this server may execute concurrently. |
| 364 | # |
| 365 | # rpc_min_threads: 16 |
| 366 | # rpc_max_threads: 2048 |
| 367 | |
| 368 | # uncomment to set socket buffer sizes on rpc connections |
| 369 | # rpc_send_buff_size_in_bytes: |
| 370 | # rpc_recv_buff_size_in_bytes: |
| 371 | |
| 372 | # Frame size for thrift (maximum field length). |
| 373 | thrift_framed_transport_size_in_mb: 15 |
| 374 | |
| 375 | # The max length of a thrift message, including all fields and |
| 376 | # internal thrift overhead. |
| 377 | thrift_max_message_length_in_mb: 16 |
| 378 | |
| 379 | # Set to true to have Cassandra create a hard link to each sstable |
| 380 | # flushed or streamed locally in a backups/ subdirectory of the |
| 381 | # Keyspace data. Removing these links is the operator's |
| 382 | # responsibility. |
| 383 | incremental_backups: false |
| 384 | |
| 385 | # Whether or not to take a snapshot before each compaction. Be |
| 386 | # careful using this option, since Cassandra won't clean up the |
| 387 | # snapshots for you. Mostly useful if you're paranoid when there |
| 388 | # is a data format change. |
| 389 | snapshot_before_compaction: false |
| 390 | |
| 391 | # Whether or not a snapshot is taken of the data before keyspace truncation |
| 392 | # or dropping of column families. The STRONGLY advised default of true |
| 393 | # should be used to provide data safety. If you set this flag to false, you will |
| 394 | # lose data on truncation or drop. |
| 395 | auto_snapshot: true |
| 396 | |
| 397 | # Add column indexes to a row after its contents reach this size. |
| 398 | # Increase if your column values are large, or if you have a very large |
| 399 | # number of columns. The competing causes are, Cassandra has to |
| 400 | # deserialize this much of the row to read a single column, so you want |
| 401 | # it to be small - at least if you do many partial-row reads - but all |
| 402 | # the index data is read for each access, so you don't want to generate |
| 403 | # that wastefully either. |
| 404 | column_index_size_in_kb: 64 |
| 405 | |
| 406 | # Size limit for rows being compacted in memory. Larger rows will spill |
| 407 | # over to disk and use a slower two-pass compaction process. A message |
| 408 | # will be logged specifying the row key. |
| 409 | in_memory_compaction_limit_in_mb: 64 |
| 410 | |
| 411 | # Number of simultaneous compactions to allow, NOT including |
| 412 | # validation "compactions" for anti-entropy repair. Simultaneous |
| 413 | # compactions can help preserve read performance in a mixed read/write |
| 414 | # workload, by mitigating the tendency of small sstables to accumulate |
| 415 | # during a single long running compactions. The default is usually |
| 416 | # fine and if you experience problems with compaction running too |
| 417 | # slowly or too fast, you should look at |
| 418 | # compaction_throughput_mb_per_sec first. |
| 419 | # |
| 420 | # concurrent_compactors defaults to the number of cores. |
| 421 | # Uncomment to make compaction mono-threaded, the pre-0.8 default. |
| 422 | #concurrent_compactors: 1 |
| 423 | |
| 424 | # Multi-threaded compaction. When enabled, each compaction will use |
| 425 | # up to one thread per core, plus one thread per sstable being merged. |
| 426 | # This is usually only useful for SSD-based hardware: otherwise, |
| 427 | # your concern is usually to get compaction to do LESS i/o (see: |
| 428 | # compaction_throughput_mb_per_sec), not more. |
| 429 | multithreaded_compaction: false |
| 430 | |
| 431 | # Throttles compaction to the given total throughput across the entire |
| 432 | # system. The faster you insert data, the faster you need to compact in |
| 433 | # order to keep the sstable count down, but in general, setting this to |
| 434 | # 16 to 32 times the rate you are inserting data is more than sufficient. |
| 435 | # Setting this to 0 disables throttling. Note that this account for all types |
| 436 | # of compaction, including validation compaction. |
| 437 | compaction_throughput_mb_per_sec: 16 |
| 438 | |
| 439 | # Track cached row keys during compaction, and re-cache their new |
| 440 | # positions in the compacted sstable. Disable if you use really large |
| 441 | # key caches. |
| 442 | compaction_preheat_key_cache: true |
| 443 | |
| 444 | # Throttles all outbound streaming file transfers on this node to the |
| 445 | # given total throughput in Mbps. This is necessary because Cassandra does |
| 446 | # mostly sequential IO when streaming data during bootstrap or repair, which |
| 447 | # can lead to saturating the network connection and degrading rpc performance. |
| 448 | # When unset, the default is 400 Mbps or 50 MB/s. |
| 449 | # stream_throughput_outbound_megabits_per_sec: 400 |
| 450 | |
| 451 | # How long the coordinator should wait for read operations to complete |
| 452 | read_request_timeout_in_ms: 10000 |
| 453 | # How long the coordinator should wait for seq or index scans to complete |
| 454 | range_request_timeout_in_ms: 10000 |
| 455 | # How long the coordinator should wait for writes to complete |
| 456 | write_request_timeout_in_ms: 10000 |
| 457 | # How long the coordinator should wait for truncates to complete |
| 458 | # (This can be much longer, because unless auto_snapshot is disabled |
| 459 | # we need to flush first so we can snapshot before removing the data.) |
| 460 | truncate_request_timeout_in_ms: 60000 |
| 461 | # The default timeout for other, miscellaneous operations |
| 462 | request_timeout_in_ms: 10000 |
| 463 | |
| 464 | # Enable operation timeout information exchange between nodes to accurately |
| 465 | # measure request timeouts, If disabled cassandra will assuming the request |
| 466 | # was forwarded to the replica instantly by the coordinator |
| 467 | # |
| 468 | # Warning: before enabling this property make sure to ntp is installed |
| 469 | # and the times are synchronized between the nodes. |
| 470 | cross_node_timeout: false |
| 471 | |
| 472 | # Enable socket timeout for streaming operation. |
| 473 | # When a timeout occurs during streaming, streaming is retried from the start |
| 474 | # of the current file. This *can* involve re-streaming an important amount of |
| 475 | # data, so you should avoid setting the value too low. |
| 476 | # Default value is 0, which never timeout streams. |
| 477 | # streaming_socket_timeout_in_ms: 0 |
| 478 | |
| 479 | # phi value that must be reached for a host to be marked down. |
| 480 | # most users should never need to adjust this. |
| 481 | # phi_convict_threshold: 8 |
| 482 | |
| 483 | # endpoint_snitch -- Set this to a class that implements |
| 484 | # IEndpointSnitch. The snitch has two functions: |
| 485 | # - it teaches Cassandra enough about your network topology to route |
| 486 | # requests efficiently |
| 487 | # - it allows Cassandra to spread replicas around your cluster to avoid |
| 488 | # correlated failures. It does this by grouping machines into |
| 489 | # "datacenters" and "racks." Cassandra will do its best not to have |
| 490 | # more than one replica on the same "rack" (which may not actually |
| 491 | # be a physical location) |
| 492 | # |
| 493 | # IF YOU CHANGE THE SNITCH AFTER DATA IS INSERTED INTO THE CLUSTER, |
| 494 | # YOU MUST RUN A FULL REPAIR, SINCE THE SNITCH AFFECTS WHERE REPLICAS |
| 495 | # ARE PLACED. |
| 496 | # |
| 497 | # Out of the box, Cassandra provides |
| 498 | # - SimpleSnitch: |
| 499 | # Treats Strategy order as proximity. This improves cache locality |
| 500 | # when disabling read repair, which can further improve throughput. |
| 501 | # Only appropriate for single-datacenter deployments. |
| 502 | # - PropertyFileSnitch: |
| 503 | # Proximity is determined by rack and data center, which are |
| 504 | # explicitly configured in cassandra-topology.properties. |
| 505 | # - GossipingPropertyFileSnitch |
| 506 | # The rack and datacenter for the local node are defined in |
| 507 | # cassandra-rackdc.properties and propagated to other nodes via gossip. If |
| 508 | # cassandra-topology.properties exists, it is used as a fallback, allowing |
| 509 | # migration from the PropertyFileSnitch. |
| 510 | # - RackInferringSnitch: |
| 511 | # Proximity is determined by rack and data center, which are |
| 512 | # assumed to correspond to the 3rd and 2nd octet of each node's |
| 513 | # IP address, respectively. Unless this happens to match your |
| 514 | # deployment conventions (as it did Facebook's), this is best used |
| 515 | # as an example of writing a custom Snitch class. |
| 516 | # - Ec2Snitch: |
| 517 | # Appropriate for EC2 deployments in a single Region. Loads Region |
| 518 | # and Availability Zone information from the EC2 API. The Region is |
| 519 | # treated as the Datacenter, and the Availability Zone as the rack. |
| 520 | # Only private IPs are used, so this will not work across multiple |
| 521 | # Regions. |
| 522 | # - Ec2MultiRegionSnitch: |
| 523 | # Uses public IPs as broadcast_address to allow cross-region |
| 524 | # connectivity. (Thus, you should set seed addresses to the public |
| 525 | # IP as well.) You will need to open the storage_port or |
| 526 | # ssl_storage_port on the public IP firewall. (For intra-Region |
| 527 | # traffic, Cassandra will switch to the private IP after |
| 528 | # establishing a connection.) |
| 529 | # |
| 530 | # You can use a custom Snitch by setting this to the full class name |
| 531 | # of the snitch, which will be assumed to be on your classpath. |
| 532 | endpoint_snitch: SimpleSnitch |
| 533 | |
| 534 | # controls how often to perform the more expensive part of host score |
| 535 | # calculation |
| 536 | dynamic_snitch_update_interval_in_ms: 100 |
| 537 | # controls how often to reset all host scores, allowing a bad host to |
| 538 | # possibly recover |
| 539 | dynamic_snitch_reset_interval_in_ms: 600000 |
| 540 | # if set greater than zero and read_repair_chance is < 1.0, this will allow |
| 541 | # 'pinning' of replicas to hosts in order to increase cache capacity. |
| 542 | # The badness threshold will control how much worse the pinned host has to be |
| 543 | # before the dynamic snitch will prefer other replicas over it. This is |
| 544 | # expressed as a double which represents a percentage. Thus, a value of |
| 545 | # 0.2 means Cassandra would continue to prefer the static snitch values |
| 546 | # until the pinned host was 20% worse than the fastest. |
| 547 | dynamic_snitch_badness_threshold: 0.1 |
| 548 | |
| 549 | # request_scheduler -- Set this to a class that implements |
| 550 | # RequestScheduler, which will schedule incoming client requests |
| 551 | # according to the specific policy. This is useful for multi-tenancy |
| 552 | # with a single Cassandra cluster. |
| 553 | # NOTE: This is specifically for requests from the client and does |
| 554 | # not affect inter node communication. |
| 555 | # org.apache.cassandra.scheduler.NoScheduler - No scheduling takes place |
| 556 | # org.apache.cassandra.scheduler.RoundRobinScheduler - Round robin of |
| 557 | # client requests to a node with a separate queue for each |
| 558 | # request_scheduler_id. The scheduler is further customized by |
| 559 | # request_scheduler_options as described below. |
| 560 | request_scheduler: org.apache.cassandra.scheduler.NoScheduler |
| 561 | |
| 562 | # Scheduler Options vary based on the type of scheduler |
| 563 | # NoScheduler - Has no options |
| 564 | # RoundRobin |
| 565 | # - throttle_limit -- The throttle_limit is the number of in-flight |
| 566 | # requests per client. Requests beyond |
| 567 | # that limit are queued up until |
| 568 | # running requests can complete. |
| 569 | # The value of 80 here is twice the number of |
| 570 | # concurrent_reads + concurrent_writes. |
| 571 | # - default_weight -- default_weight is optional and allows for |
| 572 | # overriding the default which is 1. |
| 573 | # - weights -- Weights are optional and will default to 1 or the |
| 574 | # overridden default_weight. The weight translates into how |
| 575 | # many requests are handled during each turn of the |
| 576 | # RoundRobin, based on the scheduler id. |
| 577 | # |
| 578 | # request_scheduler_options: |
| 579 | # throttle_limit: 80 |
| 580 | # default_weight: 5 |
| 581 | # weights: |
| 582 | # Keyspace1: 1 |
| 583 | # Keyspace2: 5 |
| 584 | |
| 585 | # request_scheduler_id -- An identifer based on which to perform |
| 586 | # the request scheduling. Currently the only valid option is keyspace. |
| 587 | # request_scheduler_id: keyspace |
| 588 | |
| 589 | # index_interval controls the sampling of entries from the primrary |
| 590 | # row index in terms of space versus time. The larger the interval, |
| 591 | # the smaller and less effective the sampling will be. In technicial |
| 592 | # terms, the interval coresponds to the number of index entries that |
| 593 | # are skipped between taking each sample. All the sampled entries |
| 594 | # must fit in memory. Generally, a value between 128 and 512 here |
| 595 | # coupled with a large key cache size on CFs results in the best trade |
| 596 | # offs. This value is not often changed, however if you have many |
| 597 | # very small rows (many to an OS page), then increasing this will |
| 598 | # often lower memory usage without a impact on performance. |
| 599 | index_interval: 128 |
| 600 | |
| 601 | # Enable or disable inter-node encryption |
| 602 | # Default settings are TLS v1, RSA 1024-bit keys (it is imperative that |
| 603 | # users generate their own keys) TLS_RSA_WITH_AES_128_CBC_SHA as the cipher |
| 604 | # suite for authentication, key exchange and encryption of the actual data transfers. |
| 605 | # NOTE: No custom encryption options are enabled at the moment |
| 606 | # The available internode options are : all, none, dc, rack |
| 607 | # |
| 608 | # If set to dc cassandra will encrypt the traffic between the DCs |
| 609 | # If set to rack cassandra will encrypt the traffic between the racks |
| 610 | # |
| 611 | # The passwords used in these options must match the passwords used when generating |
| 612 | # the keystore and truststore. For instructions on generating these files, see: |
| 613 | # http://download.oracle.com/javase/6/docs/technotes/guides/security/jsse/JSSERefGuide.html#CreateKeystore |
| 614 | # |
| 615 | server_encryption_options: |
| 616 | internode_encryption: none |
| 617 | keystore: conf/.keystore |
| 618 | keystore_password: cassandra |
| 619 | truststore: conf/.truststore |
| 620 | truststore_password: cassandra |
| 621 | # More advanced defaults below: |
| 622 | # protocol: TLS |
| 623 | # algorithm: SunX509 |
| 624 | # store_type: JKS |
| 625 | # cipher_suites: [TLS_RSA_WITH_AES_128_CBC_SHA,TLS_RSA_WITH_AES_256_CBC_SHA] |
| 626 | |
| 627 | # enable or disable client/server encryption. |
| 628 | client_encryption_options: |
| 629 | enabled: false |
| 630 | keystore: conf/.keystore |
| 631 | keystore_password: cassandra |
| 632 | # More advanced defaults below: |
| 633 | # protocol: TLS |
| 634 | # algorithm: SunX509 |
| 635 | # store_type: JKS |
| 636 | # cipher_suites: [TLS_RSA_WITH_AES_128_CBC_SHA,TLS_RSA_WITH_AES_256_CBC_SHA] |
| 637 | |
| 638 | # internode_compression controls whether traffic between nodes is |
| 639 | # compressed. |
| 640 | # can be: all - all traffic is compressed |
| 641 | # dc - traffic between different datacenters is compressed |
| 642 | # none - nothing is compressed. |
| 643 | internode_compression: all |