Woojoong Kim | 06e455a | 2018-01-18 13:51:46 -0800 | [diff] [blame] | 1 | # Copyright 2017-present Open Networking Foundation |
| 2 | # |
| 3 | # Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | # you may not use this file except in compliance with the License. |
| 5 | # You may obtain a copy of the License at |
| 6 | # |
| 7 | # http://www.apache.org/licenses/LICENSE-2.0 |
| 8 | # |
| 9 | # Unless required by applicable law or agreed to in writing, software |
| 10 | # distributed under the License is distributed on an "AS IS" BASIS, |
| 11 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 12 | # See the License for the specific language governing permissions and |
| 13 | # limitations under the License. |
| 14 | |
| 15 | # Cassandra storage config YAML |
| 16 | |
| 17 | # NOTE: |
| 18 | # See http://wiki.apache.org/cassandra/StorageConfiguration for |
| 19 | # full explanations of configuration directives |
| 20 | # /NOTE |
| 21 | |
| 22 | # The name of the cluster. This is mainly used to prevent machines in |
| 23 | # one logical cluster from joining another. |
| 24 | cluster_name: 'HSS Cluster' |
| 25 | |
| 26 | # This defines the number of tokens randomly assigned to this node on the ring |
| 27 | # The more tokens, relative to other nodes, the larger the proportion of data |
| 28 | # that this node will store. You probably want all nodes to have the same number |
| 29 | # of tokens assuming they have equal hardware capability. |
| 30 | # |
| 31 | # If you leave this unspecified, Cassandra will use the default of 1 token for legacy compatibility, |
| 32 | # and will use the initial_token as described below. |
| 33 | # |
| 34 | # Specifying initial_token will override this setting on the node's initial start, |
| 35 | # on subsequent starts, this setting will apply even if initial token is set. |
| 36 | # |
| 37 | # If you already have a cluster with 1 token per node, and wish to migrate to |
| 38 | # multiple tokens per node, see http://wiki.apache.org/cassandra/Operations |
| 39 | num_tokens: 256 |
| 40 | |
| 41 | # initial_token allows you to specify tokens manually. While you can use # it with |
| 42 | # vnodes (num_tokens > 1, above) -- in which case you should provide a |
| 43 | # comma-separated list -- it's primarily used when adding nodes # to legacy clusters |
| 44 | # that do not have vnodes enabled. |
| 45 | # initial_token: |
| 46 | |
| 47 | # See http://wiki.apache.org/cassandra/HintedHandoff |
| 48 | # May either be "true" or "false" to enable globally, or contain a list |
| 49 | # of data centers to enable per-datacenter. |
| 50 | # hinted_handoff_enabled: DC1,DC2 |
| 51 | hinted_handoff_enabled: true |
| 52 | # this defines the maximum amount of time a dead host will have hints |
| 53 | # generated. After it has been dead this long, new hints for it will not be |
| 54 | # created until it has been seen alive and gone down again. |
| 55 | max_hint_window_in_ms: 10800000 # 3 hours |
| 56 | # Maximum throttle in KBs per second, per delivery thread. This will be |
| 57 | # reduced proportionally to the number of nodes in the cluster. (If there |
| 58 | # are two nodes in the cluster, each delivery thread will use the maximum |
| 59 | # rate; if there are three, each will throttle to half of the maximum, |
| 60 | # since we expect two nodes to be delivering hints simultaneously.) |
| 61 | hinted_handoff_throttle_in_kb: 1024 |
| 62 | # Number of threads with which to deliver hints; |
| 63 | # Consider increasing this number when you have multi-dc deployments, since |
| 64 | # cross-dc handoff tends to be slower |
| 65 | max_hints_delivery_threads: 2 |
| 66 | |
| 67 | # Maximum throttle in KBs per second, total. This will be |
| 68 | # reduced proportionally to the number of nodes in the cluster. |
| 69 | batchlog_replay_throttle_in_kb: 1024 |
| 70 | |
| 71 | # Authentication backend, implementing IAuthenticator; used to identify users |
| 72 | # Out of the box, Cassandra provides org.apache.cassandra.auth.{AllowAllAuthenticator, |
| 73 | # PasswordAuthenticator}. |
| 74 | # |
| 75 | # - AllowAllAuthenticator performs no checks - set it to disable authentication. |
| 76 | # - PasswordAuthenticator relies on username/password pairs to authenticate |
| 77 | # users. It keeps usernames and hashed passwords in system_auth.credentials table. |
| 78 | # Please increase system_auth keyspace replication factor if you use this authenticator. |
| 79 | authenticator: AllowAllAuthenticator |
| 80 | |
| 81 | # Authorization backend, implementing IAuthorizer; used to limit access/provide permissions |
| 82 | # Out of the box, Cassandra provides org.apache.cassandra.auth.{AllowAllAuthorizer, |
| 83 | # CassandraAuthorizer}. |
| 84 | # |
| 85 | # - AllowAllAuthorizer allows any action to any user - set it to disable authorization. |
| 86 | # - CassandraAuthorizer stores permissions in system_auth.permissions table. Please |
| 87 | # increase system_auth keyspace replication factor if you use this authorizer. |
| 88 | authorizer: AllowAllAuthorizer |
| 89 | |
| 90 | # Validity period for permissions cache (fetching permissions can be an |
| 91 | # expensive operation depending on the authorizer, CassandraAuthorizer is |
| 92 | # one example). Defaults to 2000, set to 0 to disable. |
| 93 | # Will be disabled automatically for AllowAllAuthorizer. |
| 94 | permissions_validity_in_ms: 2000 |
| 95 | |
| 96 | # Refresh interval for permissions cache (if enabled). |
| 97 | # After this interval, cache entries become eligible for refresh. Upon next |
| 98 | # access, an async reload is scheduled and the old value returned until it |
| 99 | # completes. If permissions_validity_in_ms is non-zero, then this must be |
| 100 | # also. |
| 101 | # Defaults to the same value as permissions_validity_in_ms. |
| 102 | # permissions_update_interval_in_ms: 1000 |
| 103 | |
| 104 | # The partitioner is responsible for distributing groups of rows (by |
| 105 | # partition key) across nodes in the cluster. You should leave this |
| 106 | # alone for new clusters. The partitioner can NOT be changed without |
| 107 | # reloading all data, so when upgrading you should set this to the |
| 108 | # same partitioner you were already using. |
| 109 | # |
| 110 | # Besides Murmur3Partitioner, partitioners included for backwards |
| 111 | # compatibility include RandomPartitioner, ByteOrderedPartitioner, and |
| 112 | # OrderPreservingPartitioner. |
| 113 | # |
| 114 | partitioner: org.apache.cassandra.dht.Murmur3Partitioner |
| 115 | |
| 116 | # Directories where Cassandra should store data on disk. Cassandra |
| 117 | # will spread data evenly across them, subject to the granularity of |
| 118 | # the configured compaction strategy. |
| 119 | # If not set, the default directory is $CASSANDRA_HOME/data/data. |
| 120 | data_file_directories: |
| 121 | - /var/lib/cassandra/data |
| 122 | |
| 123 | # commit log. when running on magnetic HDD, this should be a |
| 124 | # separate spindle than the data directories. |
| 125 | # If not set, the default directory is $CASSANDRA_HOME/data/commitlog. |
| 126 | commitlog_directory: /var/lib/cassandra/commitlog |
| 127 | |
| 128 | # policy for data disk failures: |
| 129 | # die: shut down gossip and client transports and kill the JVM for any fs errors or |
| 130 | # single-sstable errors, so the node can be replaced. |
| 131 | # stop_paranoid: shut down gossip and client transports even for single-sstable errors, |
| 132 | # kill the JVM for errors during startup. |
| 133 | # stop: shut down gossip and client transports, leaving the node effectively dead, but |
| 134 | # can still be inspected via JMX, kill the JVM for errors during startup. |
| 135 | # best_effort: stop using the failed disk and respond to requests based on |
| 136 | # remaining available sstables. This means you WILL see obsolete |
| 137 | # data at CL.ONE! |
| 138 | # ignore: ignore fatal errors and let requests fail, as in pre-1.2 Cassandra |
| 139 | disk_failure_policy: stop |
| 140 | |
| 141 | # policy for commit disk failures: |
| 142 | # die: shut down gossip and Thrift and kill the JVM, so the node can be replaced. |
| 143 | # stop: shut down gossip and Thrift, leaving the node effectively dead, but |
| 144 | # can still be inspected via JMX. |
| 145 | # stop_commit: shutdown the commit log, letting writes collect but |
| 146 | # continuing to service reads, as in pre-2.0.5 Cassandra |
| 147 | # ignore: ignore fatal errors and let the batches fail |
| 148 | commit_failure_policy: stop |
| 149 | |
| 150 | # Maximum size of the key cache in memory. |
| 151 | # |
| 152 | # Each key cache hit saves 1 seek and each row cache hit saves 2 seeks at the |
| 153 | # minimum, sometimes more. The key cache is fairly tiny for the amount of |
| 154 | # time it saves, so it's worthwhile to use it at large numbers. |
| 155 | # The row cache saves even more time, but must contain the entire row, |
| 156 | # so it is extremely space-intensive. It's best to only use the |
| 157 | # row cache if you have hot rows or static rows. |
| 158 | # |
| 159 | # NOTE: if you reduce the size, you may not get you hottest keys loaded on startup. |
| 160 | # |
| 161 | # Default value is empty to make it "auto" (min(5% of Heap (in MB), 100MB)). Set to 0 to disable key cache. |
| 162 | key_cache_size_in_mb: |
| 163 | |
| 164 | # Duration in seconds after which Cassandra should |
| 165 | # save the key cache. Caches are saved to saved_caches_directory as |
| 166 | # specified in this configuration file. |
| 167 | # |
| 168 | # Saved caches greatly improve cold-start speeds, and is relatively cheap in |
| 169 | # terms of I/O for the key cache. Row cache saving is much more expensive and |
| 170 | # has limited use. |
| 171 | # |
| 172 | # Default is 14400 or 4 hours. |
| 173 | key_cache_save_period: 14400 |
| 174 | |
| 175 | # Number of keys from the key cache to save |
| 176 | # Disabled by default, meaning all keys are going to be saved |
| 177 | # key_cache_keys_to_save: 100 |
| 178 | |
| 179 | # Maximum size of the row cache in memory. |
| 180 | # NOTE: if you reduce the size, you may not get you hottest keys loaded on startup. |
| 181 | # |
| 182 | # Default value is 0, to disable row caching. |
| 183 | row_cache_size_in_mb: 0 |
| 184 | |
| 185 | # Duration in seconds after which Cassandra should |
| 186 | # save the row cache. Caches are saved to saved_caches_directory as specified |
| 187 | # in this configuration file. |
| 188 | # |
| 189 | # Saved caches greatly improve cold-start speeds, and is relatively cheap in |
| 190 | # terms of I/O for the key cache. Row cache saving is much more expensive and |
| 191 | # has limited use. |
| 192 | # |
| 193 | # Default is 0 to disable saving the row cache. |
| 194 | row_cache_save_period: 0 |
| 195 | |
| 196 | # Number of keys from the row cache to save |
| 197 | # Disabled by default, meaning all keys are going to be saved |
| 198 | # row_cache_keys_to_save: 100 |
| 199 | |
| 200 | # Maximum size of the counter cache in memory. |
| 201 | # |
| 202 | # Counter cache helps to reduce counter locks' contention for hot counter cells. |
| 203 | # In case of RF = 1 a counter cache hit will cause Cassandra to skip the read before |
| 204 | # write entirely. With RF > 1 a counter cache hit will still help to reduce the duration |
| 205 | # of the lock hold, helping with hot counter cell updates, but will not allow skipping |
| 206 | # the read entirely. Only the local (clock, count) tuple of a counter cell is kept |
| 207 | # in memory, not the whole counter, so it's relatively cheap. |
| 208 | # |
| 209 | # NOTE: if you reduce the size, you may not get you hottest keys loaded on startup. |
| 210 | # |
| 211 | # Default value is empty to make it "auto" (min(2.5% of Heap (in MB), 50MB)). Set to 0 to disable counter cache. |
| 212 | # NOTE: if you perform counter deletes and rely on low gcgs, you should disable the counter cache. |
| 213 | counter_cache_size_in_mb: |
| 214 | |
| 215 | # Duration in seconds after which Cassandra should |
| 216 | # save the counter cache (keys only). Caches are saved to saved_caches_directory as |
| 217 | # specified in this configuration file. |
| 218 | # |
| 219 | # Default is 7200 or 2 hours. |
| 220 | counter_cache_save_period: 7200 |
| 221 | |
| 222 | # Number of keys from the counter cache to save |
| 223 | # Disabled by default, meaning all keys are going to be saved |
| 224 | # counter_cache_keys_to_save: 100 |
| 225 | |
| 226 | # The off-heap memory allocator. Affects storage engine metadata as |
| 227 | # well as caches. Experiments show that JEMAlloc saves some memory |
| 228 | # than the native GCC allocator (i.e., JEMalloc is more |
| 229 | # fragmentation-resistant). |
| 230 | # |
| 231 | # Supported values are: NativeAllocator, JEMallocAllocator |
| 232 | # |
| 233 | # If you intend to use JEMallocAllocator you have to install JEMalloc as library and |
| 234 | # modify cassandra-env.sh as directed in the file. |
| 235 | # |
| 236 | # Defaults to NativeAllocator |
| 237 | # memory_allocator: NativeAllocator |
| 238 | |
| 239 | # saved caches |
| 240 | # If not set, the default directory is $CASSANDRA_HOME/data/saved_caches. |
| 241 | saved_caches_directory: /var/lib/cassandra/saved_caches |
| 242 | |
| 243 | # commitlog_sync may be either "periodic" or "batch." |
| 244 | # |
| 245 | # When in batch mode, Cassandra won't ack writes until the commit log |
| 246 | # has been fsynced to disk. It will wait |
| 247 | # commitlog_sync_batch_window_in_ms milliseconds between fsyncs. |
| 248 | # This window should be kept short because the writer threads will |
| 249 | # be unable to do extra work while waiting. (You may need to increase |
| 250 | # concurrent_writes for the same reason.) |
| 251 | # |
| 252 | # commitlog_sync: batch |
| 253 | # commitlog_sync_batch_window_in_ms: 2 |
| 254 | # |
| 255 | # the other option is "periodic" where writes may be acked immediately |
| 256 | # and the CommitLog is simply synced every commitlog_sync_period_in_ms |
| 257 | # milliseconds. |
| 258 | commitlog_sync: periodic |
| 259 | commitlog_sync_period_in_ms: 10000 |
| 260 | |
| 261 | # The size of the individual commitlog file segments. A commitlog |
| 262 | # segment may be archived, deleted, or recycled once all the data |
| 263 | # in it (potentially from each columnfamily in the system) has been |
| 264 | # flushed to sstables. |
| 265 | # |
| 266 | # The default size is 32, which is almost always fine, but if you are |
| 267 | # archiving commitlog segments (see commitlog_archiving.properties), |
| 268 | # then you probably want a finer granularity of archiving; 8 or 16 MB |
| 269 | # is reasonable. |
| 270 | commitlog_segment_size_in_mb: 32 |
| 271 | |
| 272 | # Reuse commit log files when possible. The default is false, and this |
| 273 | # feature will be removed entirely in future versions of Cassandra. |
| 274 | #commitlog_segment_recycling: false |
| 275 | |
| 276 | # any class that implements the SeedProvider interface and has a |
| 277 | # constructor that takes a Map<String, String> of parameters will do. |
| 278 | seed_provider: |
| 279 | # Addresses of hosts that are deemed contact points. |
| 280 | # Cassandra nodes use this list of hosts to find each other and learn |
| 281 | # the topology of the ring. You must change this if you are running |
| 282 | # multiple nodes! |
| 283 | - class_name: org.apache.cassandra.locator.SimpleSeedProvider |
| 284 | parameters: |
| 285 | # seeds is actually a comma-delimited list of addresses. |
| 286 | # Ex: "<ip1>,<ip2>,<ip3>" |
| 287 | - seeds: "{{ db_ip }}" |
| 288 | |
| 289 | # For workloads with more data than can fit in memory, Cassandra's |
| 290 | # bottleneck will be reads that need to fetch data from |
| 291 | # disk. "concurrent_reads" should be set to (16 * number_of_drives) in |
| 292 | # order to allow the operations to enqueue low enough in the stack |
| 293 | # that the OS and drives can reorder them. Same applies to |
| 294 | # "concurrent_counter_writes", since counter writes read the current |
| 295 | # values before incrementing and writing them back. |
| 296 | # |
| 297 | # On the other hand, since writes are almost never IO bound, the ideal |
| 298 | # number of "concurrent_writes" is dependent on the number of cores in |
| 299 | # your system; (8 * number_of_cores) is a good rule of thumb. |
| 300 | concurrent_reads: 32 |
| 301 | concurrent_writes: 32 |
| 302 | concurrent_counter_writes: 32 |
| 303 | |
| 304 | # Total memory to use for sstable-reading buffers. Defaults to |
| 305 | # the smaller of 1/4 of heap or 512MB. |
| 306 | # file_cache_size_in_mb: 512 |
| 307 | |
| 308 | # Total permitted memory to use for memtables. Cassandra will stop |
| 309 | # accepting writes when the limit is exceeded until a flush completes, |
| 310 | # and will trigger a flush based on memtable_cleanup_threshold |
| 311 | # If omitted, Cassandra will set both to 1/4 the size of the heap. |
| 312 | # memtable_heap_space_in_mb: 2048 |
| 313 | # memtable_offheap_space_in_mb: 2048 |
| 314 | |
| 315 | # Ratio of occupied non-flushing memtable size to total permitted size |
| 316 | # that will trigger a flush of the largest memtable. Lager mct will |
| 317 | # mean larger flushes and hence less compaction, but also less concurrent |
| 318 | # flush activity which can make it difficult to keep your disks fed |
| 319 | # under heavy write load. |
| 320 | # |
| 321 | # memtable_cleanup_threshold defaults to 1 / (memtable_flush_writers + 1) |
| 322 | # memtable_cleanup_threshold: 0.11 |
| 323 | |
| 324 | # Specify the way Cassandra allocates and manages memtable memory. |
| 325 | # Options are: |
| 326 | # heap_buffers: on heap nio buffers |
| 327 | # offheap_buffers: off heap (direct) nio buffers |
| 328 | # offheap_objects: native memory, eliminating nio buffer heap overhead |
| 329 | memtable_allocation_type: heap_buffers |
| 330 | |
| 331 | # Total space to use for commitlogs. Since commitlog segments are |
| 332 | # mmapped, and hence use up address space, the default size is 32 |
| 333 | # on 32-bit JVMs, and 8192 on 64-bit JVMs. |
| 334 | # |
| 335 | # If space gets above this value (it will round up to the next nearest |
| 336 | # segment multiple), Cassandra will flush every dirty CF in the oldest |
| 337 | # segment and remove it. So a small total commitlog space will tend |
| 338 | # to cause more flush activity on less-active columnfamilies. |
| 339 | # commitlog_total_space_in_mb: 8192 |
| 340 | |
| 341 | # This sets the amount of memtable flush writer threads. These will |
| 342 | # be blocked by disk io, and each one will hold a memtable in memory |
| 343 | # while blocked. |
| 344 | # |
| 345 | # memtable_flush_writers defaults to the smaller of (number of disks, |
| 346 | # number of cores), with a minimum of 2 and a maximum of 8. |
| 347 | # |
| 348 | # If your data directories are backed by SSD, you should increase this |
| 349 | # to the number of cores. |
| 350 | #memtable_flush_writers: 8 |
| 351 | |
| 352 | # A fixed memory pool size in MB for for SSTable index summaries. If left |
| 353 | # empty, this will default to 5% of the heap size. If the memory usage of |
| 354 | # all index summaries exceeds this limit, SSTables with low read rates will |
| 355 | # shrink their index summaries in order to meet this limit. However, this |
| 356 | # is a best-effort process. In extreme conditions Cassandra may need to use |
| 357 | # more than this amount of memory. |
| 358 | index_summary_capacity_in_mb: |
| 359 | |
| 360 | # How frequently index summaries should be resampled. This is done |
| 361 | # periodically to redistribute memory from the fixed-size pool to sstables |
| 362 | # proportional their recent read rates. Setting to -1 will disable this |
| 363 | # process, leaving existing index summaries at their current sampling level. |
| 364 | index_summary_resize_interval_in_minutes: 60 |
| 365 | |
| 366 | # Whether to, when doing sequential writing, fsync() at intervals in |
| 367 | # order to force the operating system to flush the dirty |
| 368 | # buffers. Enable this to avoid sudden dirty buffer flushing from |
| 369 | # impacting read latencies. Almost always a good idea on SSDs; not |
| 370 | # necessarily on platters. |
| 371 | trickle_fsync: false |
| 372 | trickle_fsync_interval_in_kb: 10240 |
| 373 | |
| 374 | # TCP port, for commands and data |
| 375 | # For security reasons, you should not expose this port to the internet. Firewall it if needed. |
| 376 | storage_port: 7000 |
| 377 | |
| 378 | # SSL port, for encrypted communication. Unused unless enabled in |
| 379 | # encryption_options |
| 380 | # For security reasons, you should not expose this port to the internet. Firewall it if needed. |
| 381 | ssl_storage_port: 7001 |
| 382 | |
| 383 | # Address or interface to bind to and tell other Cassandra nodes to connect to. |
| 384 | # You _must_ change this if you want multiple nodes to be able to communicate! |
| 385 | # |
| 386 | # Set listen_address OR listen_interface, not both. Interfaces must correspond |
| 387 | # to a single address, IP aliasing is not supported. |
| 388 | # |
| 389 | # Leaving it blank leaves it up to InetAddress.getLocalHost(). This |
| 390 | # will always do the Right Thing _if_ the node is properly configured |
| 391 | # (hostname, name resolution, etc), and the Right Thing is to use the |
| 392 | # address associated with the hostname (it might not be). |
| 393 | # |
| 394 | # Setting listen_address to 0.0.0.0 is always wrong. |
| 395 | # |
| 396 | # If you choose to specify the interface by name and the interface has an ipv4 and an ipv6 address |
| 397 | # you can specify which should be chosen using listen_interface_prefer_ipv6. If false the first ipv4 |
| 398 | # address will be used. If true the first ipv6 address will be used. Defaults to false preferring |
| 399 | # ipv4. If there is only one address it will be selected regardless of ipv4/ipv6. |
| 400 | listen_address: {{ db_ip }} |
| 401 | # listen_interface: eth0 |
| 402 | # listen_interface_prefer_ipv6: false |
| 403 | |
| 404 | # Address to broadcast to other Cassandra nodes |
| 405 | # Leaving this blank will set it to the same value as listen_address |
| 406 | #broadcast_address: 192.168.112.60 |
| 407 | |
| 408 | # Internode authentication backend, implementing IInternodeAuthenticator; |
| 409 | # used to allow/disallow connections from peer nodes. |
| 410 | # internode_authenticator: org.apache.cassandra.auth.AllowAllInternodeAuthenticator |
| 411 | |
| 412 | # Whether to start the native transport server. |
| 413 | # Please note that the address on which the native transport is bound is the |
| 414 | # same as the rpc_address. The port however is different and specified below. |
| 415 | start_native_transport: true |
| 416 | # port for the CQL native transport to listen for clients on |
| 417 | # For security reasons, you should not expose this port to the internet. Firewall it if needed. |
| 418 | native_transport_port: 9042 |
| 419 | # The maximum threads for handling requests when the native transport is used. |
| 420 | # This is similar to rpc_max_threads though the default differs slightly (and |
| 421 | # there is no native_transport_min_threads, idle threads will always be stopped |
| 422 | # after 30 seconds). |
| 423 | # native_transport_max_threads: 128 |
| 424 | # |
| 425 | # The maximum size of allowed frame. Frame (requests) larger than this will |
| 426 | # be rejected as invalid. The default is 256MB. |
| 427 | # native_transport_max_frame_size_in_mb: 256 |
| 428 | |
| 429 | # The maximum number of concurrent client connections. |
| 430 | # The default is -1, which means unlimited. |
| 431 | # native_transport_max_concurrent_connections: -1 |
| 432 | |
| 433 | # The maximum number of concurrent client connections per source ip. |
| 434 | # The default is -1, which means unlimited. |
| 435 | # native_transport_max_concurrent_connections_per_ip: -1 |
| 436 | |
| 437 | # Whether to start the thrift rpc server. |
| 438 | start_rpc: true |
| 439 | |
| 440 | # The address or interface to bind the Thrift RPC service and native transport |
| 441 | # server to. |
| 442 | # |
| 443 | # Set rpc_address OR rpc_interface, not both. Interfaces must correspond |
| 444 | # to a single address, IP aliasing is not supported. |
| 445 | # |
| 446 | # Leaving rpc_address blank has the same effect as on listen_address |
| 447 | # (i.e. it will be based on the configured hostname of the node). |
| 448 | # |
| 449 | # Note that unlike listen_address, you can specify 0.0.0.0, but you must also |
| 450 | # set broadcast_rpc_address to a value other than 0.0.0.0. |
| 451 | # |
| 452 | # For security reasons, you should not expose this port to the internet. Firewall it if needed. |
| 453 | # |
| 454 | # If you choose to specify the interface by name and the interface has an ipv4 and an ipv6 address |
| 455 | # you can specify which should be chosen using rpc_interface_prefer_ipv6. If false the first ipv4 |
| 456 | # address will be used. If true the first ipv6 address will be used. Defaults to false preferring |
| 457 | # ipv4. If there is only one address it will be selected regardless of ipv4/ipv6. |
| 458 | rpc_address: {{ db_ip }} |
| 459 | # rpc_interface: eth1 |
| 460 | # rpc_interface_prefer_ipv6: false |
| 461 | |
| 462 | # port for Thrift to listen for clients on |
| 463 | rpc_port: 9160 |
| 464 | |
| 465 | # RPC address to broadcast to drivers and other Cassandra nodes. This cannot |
| 466 | # be set to 0.0.0.0. If left blank, this will be set to the value of |
| 467 | # rpc_address. If rpc_address is set to 0.0.0.0, broadcast_rpc_address must |
| 468 | # be set. |
| 469 | # broadcast_rpc_address: 1.2.3.4 |
| 470 | |
| 471 | # enable or disable keepalive on rpc/native connections |
| 472 | rpc_keepalive: true |
| 473 | |
| 474 | # Cassandra provides two out-of-the-box options for the RPC Server: |
| 475 | # |
| 476 | # sync -> One thread per thrift connection. For a very large number of clients, memory |
| 477 | # will be your limiting factor. On a 64 bit JVM, 180KB is the minimum stack size |
| 478 | # per thread, and that will correspond to your use of virtual memory (but physical memory |
| 479 | # may be limited depending on use of stack space). |
| 480 | # |
| 481 | # hsha -> Stands for "half synchronous, half asynchronous." All thrift clients are handled |
| 482 | # asynchronously using a small number of threads that does not vary with the amount |
| 483 | # of thrift clients (and thus scales well to many clients). The rpc requests are still |
| 484 | # synchronous (one thread per active request). If hsha is selected then it is essential |
| 485 | # that rpc_max_threads is changed from the default value of unlimited. |
| 486 | # |
| 487 | # The default is sync because on Windows hsha is about 30% slower. On Linux, |
| 488 | # sync/hsha performance is about the same, with hsha of course using less memory. |
| 489 | # |
| 490 | # Alternatively, can provide your own RPC server by providing the fully-qualified class name |
| 491 | # of an o.a.c.t.TServerFactory that can create an instance of it. |
| 492 | rpc_server_type: sync |
| 493 | |
| 494 | # Uncomment rpc_min|max_thread to set request pool size limits. |
| 495 | # |
| 496 | # Regardless of your choice of RPC server (see above), the number of maximum requests in the |
| 497 | # RPC thread pool dictates how many concurrent requests are possible (but if you are using the sync |
| 498 | # RPC server, it also dictates the number of clients that can be connected at all). |
| 499 | # |
| 500 | # The default is unlimited and thus provides no protection against clients overwhelming the server. You are |
| 501 | # encouraged to set a maximum that makes sense for you in production, but do keep in mind that |
| 502 | # rpc_max_threads represents the maximum number of client requests this server may execute concurrently. |
| 503 | # |
| 504 | # rpc_min_threads: 16 |
| 505 | # rpc_max_threads: 2048 |
| 506 | |
| 507 | # uncomment to set socket buffer sizes on rpc connections |
| 508 | # rpc_send_buff_size_in_bytes: |
| 509 | # rpc_recv_buff_size_in_bytes: |
| 510 | |
| 511 | # Uncomment to set socket buffer size for internode communication |
| 512 | # Note that when setting this, the buffer size is limited by net.core.wmem_max |
| 513 | # and when not setting it it is defined by net.ipv4.tcp_wmem |
| 514 | # See: |
| 515 | # /proc/sys/net/core/wmem_max |
| 516 | # /proc/sys/net/core/rmem_max |
| 517 | # /proc/sys/net/ipv4/tcp_wmem |
| 518 | # /proc/sys/net/ipv4/tcp_wmem |
| 519 | # and: man tcp |
| 520 | # internode_send_buff_size_in_bytes: |
| 521 | # internode_recv_buff_size_in_bytes: |
| 522 | |
| 523 | # Frame size for thrift (maximum message length). |
| 524 | thrift_framed_transport_size_in_mb: 15 |
| 525 | |
| 526 | # Set to true to have Cassandra create a hard link to each sstable |
| 527 | # flushed or streamed locally in a backups/ subdirectory of the |
| 528 | # keyspace data. Removing these links is the operator's |
| 529 | # responsibility. |
| 530 | incremental_backups: false |
| 531 | |
| 532 | # Whether or not to take a snapshot before each compaction. Be |
| 533 | # careful using this option, since Cassandra won't clean up the |
| 534 | # snapshots for you. Mostly useful if you're paranoid when there |
| 535 | # is a data format change. |
| 536 | snapshot_before_compaction: false |
| 537 | |
| 538 | # Whether or not a snapshot is taken of the data before keyspace truncation |
| 539 | # or dropping of column families. The STRONGLY advised default of true |
| 540 | # should be used to provide data safety. If you set this flag to false, you will |
| 541 | # lose data on truncation or drop. |
| 542 | auto_snapshot: true |
| 543 | |
| 544 | # When executing a scan, within or across a partition, we need to keep the |
| 545 | # tombstones seen in memory so we can return them to the coordinator, which |
| 546 | # will use them to make sure other replicas also know about the deleted rows. |
| 547 | # With workloads that generate a lot of tombstones, this can cause performance |
| 548 | # problems and even exaust the server heap. |
| 549 | # (http://www.datastax.com/dev/blog/cassandra-anti-patterns-queues-and-queue-like-datasets) |
| 550 | # Adjust the thresholds here if you understand the dangers and want to |
| 551 | # scan more tombstones anyway. These thresholds may also be adjusted at runtime |
| 552 | # using the StorageService mbean. |
| 553 | tombstone_warn_threshold: 1000 |
| 554 | tombstone_failure_threshold: 100000 |
| 555 | |
| 556 | # Granularity of the collation index of rows within a partition. |
| 557 | # Increase if your rows are large, or if you have a very large |
| 558 | # number of rows per partition. The competing goals are these: |
| 559 | # 1) a smaller granularity means more index entries are generated |
| 560 | # and looking up rows withing the partition by collation column |
| 561 | # is faster |
| 562 | # 2) but, Cassandra will keep the collation index in memory for hot |
| 563 | # rows (as part of the key cache), so a larger granularity means |
| 564 | # you can cache more hot rows |
| 565 | column_index_size_in_kb: 64 |
| 566 | |
| 567 | |
| 568 | # Log WARN on any batch size exceeding this value. 5kb per batch by default. |
| 569 | # Caution should be taken on increasing the size of this threshold as it can lead to node instability. |
| 570 | batch_size_warn_threshold_in_kb: 5 |
| 571 | |
| 572 | |
| 573 | # Log WARN on any batches not of type LOGGED than span across more partitions than this limit |
| 574 | unlogged_batch_across_partitions_warn_threshold: 10 |
| 575 | |
| 576 | # Number of simultaneous compactions to allow, NOT including |
| 577 | # validation "compactions" for anti-entropy repair. Simultaneous |
| 578 | # compactions can help preserve read performance in a mixed read/write |
| 579 | # workload, by mitigating the tendency of small sstables to accumulate |
| 580 | # during a single long running compactions. The default is usually |
| 581 | # fine and if you experience problems with compaction running too |
| 582 | # slowly or too fast, you should look at |
| 583 | # compaction_throughput_mb_per_sec first. |
| 584 | # |
| 585 | # concurrent_compactors defaults to the smaller of (number of disks, |
| 586 | # number of cores), with a minimum of 2 and a maximum of 8. |
| 587 | # |
| 588 | # If your data directories are backed by SSD, you should increase this |
| 589 | # to the number of cores. |
| 590 | #concurrent_compactors: 1 |
| 591 | |
| 592 | # Throttles compaction to the given total throughput across the entire |
| 593 | # system. The faster you insert data, the faster you need to compact in |
| 594 | # order to keep the sstable count down, but in general, setting this to |
| 595 | # 16 to 32 times the rate you are inserting data is more than sufficient. |
| 596 | # Setting this to 0 disables throttling. Note that this account for all types |
| 597 | # of compaction, including validation compaction. |
| 598 | compaction_throughput_mb_per_sec: 16 |
| 599 | |
| 600 | # Log a warning when compacting partitions larger than this value |
| 601 | compaction_large_partition_warning_threshold_mb: 100 |
| 602 | |
| 603 | # When compacting, the replacement sstable(s) can be opened before they |
| 604 | # are completely written, and used in place of the prior sstables for |
| 605 | # any range that has been written. This helps to smoothly transfer reads |
| 606 | # between the sstables, reducing page cache churn and keeping hot rows hot |
| 607 | sstable_preemptive_open_interval_in_mb: 50 |
| 608 | |
| 609 | # Throttles all outbound streaming file transfers on this node to the |
| 610 | # given total throughput in Mbps. This is necessary because Cassandra does |
| 611 | # mostly sequential IO when streaming data during bootstrap or repair, which |
| 612 | # can lead to saturating the network connection and degrading rpc performance. |
| 613 | # When unset, the default is 200 Mbps or 25 MB/s. |
| 614 | # stream_throughput_outbound_megabits_per_sec: 200 |
| 615 | |
| 616 | # Throttles all streaming file transfer between the datacenters, |
| 617 | # this setting allows users to throttle inter dc stream throughput in addition |
| 618 | # to throttling all network stream traffic as configured with |
| 619 | # stream_throughput_outbound_megabits_per_sec |
| 620 | # When unset, the default is 200 Mbps or 25 MB/s |
| 621 | # inter_dc_stream_throughput_outbound_megabits_per_sec: 200 |
| 622 | |
| 623 | # How long the coordinator should wait for read operations to complete |
| 624 | read_request_timeout_in_ms: 5000 |
| 625 | # How long the coordinator should wait for seq or index scans to complete |
| 626 | range_request_timeout_in_ms: 10000 |
| 627 | # How long the coordinator should wait for writes to complete |
| 628 | write_request_timeout_in_ms: 2000 |
| 629 | # How long the coordinator should wait for counter writes to complete |
| 630 | counter_write_request_timeout_in_ms: 5000 |
| 631 | # How long a coordinator should continue to retry a CAS operation |
| 632 | # that contends with other proposals for the same row |
| 633 | cas_contention_timeout_in_ms: 1000 |
| 634 | # How long the coordinator should wait for truncates to complete |
| 635 | # (This can be much longer, because unless auto_snapshot is disabled |
| 636 | # we need to flush first so we can snapshot before removing the data.) |
| 637 | truncate_request_timeout_in_ms: 60000 |
| 638 | # The default timeout for other, miscellaneous operations |
| 639 | request_timeout_in_ms: 10000 |
| 640 | |
| 641 | # Enable operation timeout information exchange between nodes to accurately |
| 642 | # measure request timeouts. If disabled, replicas will assume that requests |
| 643 | # were forwarded to them instantly by the coordinator, which means that |
| 644 | # under overload conditions we will waste that much extra time processing |
| 645 | # already-timed-out requests. |
| 646 | # |
| 647 | # Warning: before enabling this property make sure to ntp is installed |
| 648 | # and the times are synchronized between the nodes. |
| 649 | cross_node_timeout: false |
| 650 | |
| 651 | # Set socket timeout for streaming operation. |
| 652 | # The stream session is failed if no data/ack is received by any of the participants |
| 653 | # within that period, which means this should also be sufficient to stream a large |
| 654 | # sstable or rebuild table indexes. |
| 655 | # Default value is 86400000ms, which means stale streams timeout after 24 hours. |
| 656 | # A value of zero means stream sockets should never time out. |
| 657 | # streaming_socket_timeout_in_ms: 86400000 |
| 658 | |
| 659 | # phi value that must be reached for a host to be marked down. |
| 660 | # most users should never need to adjust this. |
| 661 | # phi_convict_threshold: 8 |
| 662 | |
| 663 | # endpoint_snitch -- Set this to a class that implements |
| 664 | # IEndpointSnitch. The snitch has two functions: |
| 665 | # - it teaches Cassandra enough about your network topology to route |
| 666 | # requests efficiently |
| 667 | # - it allows Cassandra to spread replicas around your cluster to avoid |
| 668 | # correlated failures. It does this by grouping machines into |
| 669 | # "datacenters" and "racks." Cassandra will do its best not to have |
| 670 | # more than one replica on the same "rack" (which may not actually |
| 671 | # be a physical location) |
| 672 | # |
| 673 | # CASSANDRA WILL NOT ALLOW YOU TO SWITCH TO AN INCOMPATIBLE SNITCH |
| 674 | # ONCE DATA IS INSERTED INTO THE CLUSTER. This would cause data loss. |
| 675 | # This means that if you start with the default SimpleSnitch, which |
| 676 | # locates every node on "rack1" in "datacenter1", your only options |
| 677 | # if you need to add another datacenter are GossipingPropertyFileSnitch |
| 678 | # (and the older PFS). From there, if you want to migrate to an |
| 679 | # incompatible snitch like Ec2Snitch you can do it by adding new nodes |
| 680 | # under Ec2Snitch (which will locate them in a new "datacenter") and |
| 681 | # decommissioning the old ones. |
| 682 | # |
| 683 | # Out of the box, Cassandra provides |
| 684 | # - SimpleSnitch: |
| 685 | # Treats Strategy order as proximity. This can improve cache |
| 686 | # locality when disabling read repair. Only appropriate for |
| 687 | # single-datacenter deployments. |
| 688 | # - GossipingPropertyFileSnitch |
| 689 | # This should be your go-to snitch for production use. The rack |
| 690 | # and datacenter for the local node are defined in |
| 691 | # cassandra-rackdc.properties and propagated to other nodes via |
| 692 | # gossip. If cassandra-topology.properties exists, it is used as a |
| 693 | # fallback, allowing migration from the PropertyFileSnitch. |
| 694 | # - PropertyFileSnitch: |
| 695 | # Proximity is determined by rack and data center, which are |
| 696 | # explicitly configured in cassandra-topology.properties. |
| 697 | # - Ec2Snitch: |
| 698 | # Appropriate for EC2 deployments in a single Region. Loads Region |
| 699 | # and Availability Zone information from the EC2 API. The Region is |
| 700 | # treated as the datacenter, and the Availability Zone as the rack. |
| 701 | # Only private IPs are used, so this will not work across multiple |
| 702 | # Regions. |
| 703 | # - Ec2MultiRegionSnitch: |
| 704 | # Uses public IPs as broadcast_address to allow cross-region |
| 705 | # connectivity. (Thus, you should set seed addresses to the public |
| 706 | # IP as well.) You will need to open the storage_port or |
| 707 | # ssl_storage_port on the public IP firewall. (For intra-Region |
| 708 | # traffic, Cassandra will switch to the private IP after |
| 709 | # establishing a connection.) |
| 710 | # - RackInferringSnitch: |
| 711 | # Proximity is determined by rack and data center, which are |
| 712 | # assumed to correspond to the 3rd and 2nd octet of each node's IP |
| 713 | # address, respectively. Unless this happens to match your |
| 714 | # deployment conventions, this is best used as an example of |
| 715 | # writing a custom Snitch class and is provided in that spirit. |
| 716 | # |
| 717 | # You can use a custom Snitch by setting this to the full class name |
| 718 | # of the snitch, which will be assumed to be on your classpath. |
| 719 | endpoint_snitch: GossipingPropertyFileSnitch |
| 720 | |
| 721 | # controls how often to perform the more expensive part of host score |
| 722 | # calculation |
| 723 | dynamic_snitch_update_interval_in_ms: 100 |
| 724 | # controls how often to reset all host scores, allowing a bad host to |
| 725 | # possibly recover |
| 726 | dynamic_snitch_reset_interval_in_ms: 600000 |
| 727 | # if set greater than zero and read_repair_chance is < 1.0, this will allow |
| 728 | # 'pinning' of replicas to hosts in order to increase cache capacity. |
| 729 | # The badness threshold will control how much worse the pinned host has to be |
| 730 | # before the dynamic snitch will prefer other replicas over it. This is |
| 731 | # expressed as a double which represents a percentage. Thus, a value of |
| 732 | # 0.2 means Cassandra would continue to prefer the static snitch values |
| 733 | # until the pinned host was 20% worse than the fastest. |
| 734 | dynamic_snitch_badness_threshold: 0.1 |
| 735 | |
| 736 | # request_scheduler -- Set this to a class that implements |
| 737 | # RequestScheduler, which will schedule incoming client requests |
| 738 | # according to the specific policy. This is useful for multi-tenancy |
| 739 | # with a single Cassandra cluster. |
| 740 | # NOTE: This is specifically for requests from the client and does |
| 741 | # not affect inter node communication. |
| 742 | # org.apache.cassandra.scheduler.NoScheduler - No scheduling takes place |
| 743 | # org.apache.cassandra.scheduler.RoundRobinScheduler - Round robin of |
| 744 | # client requests to a node with a separate queue for each |
| 745 | # request_scheduler_id. The scheduler is further customized by |
| 746 | # request_scheduler_options as described below. |
| 747 | request_scheduler: org.apache.cassandra.scheduler.NoScheduler |
| 748 | |
| 749 | # Scheduler Options vary based on the type of scheduler |
| 750 | # NoScheduler - Has no options |
| 751 | # RoundRobin |
| 752 | # - throttle_limit -- The throttle_limit is the number of in-flight |
| 753 | # requests per client. Requests beyond |
| 754 | # that limit are queued up until |
| 755 | # running requests can complete. |
| 756 | # The value of 80 here is twice the number of |
| 757 | # concurrent_reads + concurrent_writes. |
| 758 | # - default_weight -- default_weight is optional and allows for |
| 759 | # overriding the default which is 1. |
| 760 | # - weights -- Weights are optional and will default to 1 or the |
| 761 | # overridden default_weight. The weight translates into how |
| 762 | # many requests are handled during each turn of the |
| 763 | # RoundRobin, based on the scheduler id. |
| 764 | # |
| 765 | # request_scheduler_options: |
| 766 | # throttle_limit: 80 |
| 767 | # default_weight: 5 |
| 768 | # weights: |
| 769 | # Keyspace1: 1 |
| 770 | # Keyspace2: 5 |
| 771 | |
| 772 | # request_scheduler_id -- An identifier based on which to perform |
| 773 | # the request scheduling. Currently the only valid option is keyspace. |
| 774 | # request_scheduler_id: keyspace |
| 775 | |
| 776 | # Enable or disable inter-node encryption |
| 777 | # Default settings are TLS v1, RSA 1024-bit keys (it is imperative that |
| 778 | # users generate their own keys) TLS_RSA_WITH_AES_128_CBC_SHA as the cipher |
| 779 | # suite for authentication, key exchange and encryption of the actual data transfers. |
| 780 | # Use the DHE/ECDHE ciphers if running in FIPS 140 compliant mode. |
| 781 | # NOTE: No custom encryption options are enabled at the moment |
| 782 | # The available internode options are : all, none, dc, rack |
| 783 | # |
| 784 | # If set to dc cassandra will encrypt the traffic between the DCs |
| 785 | # If set to rack cassandra will encrypt the traffic between the racks |
| 786 | # |
| 787 | # The passwords used in these options must match the passwords used when generating |
| 788 | # the keystore and truststore. For instructions on generating these files, see: |
| 789 | # http://download.oracle.com/javase/6/docs/technotes/guides/security/jsse/JSSERefGuide.html#CreateKeystore |
| 790 | # |
| 791 | server_encryption_options: |
| 792 | internode_encryption: none |
| 793 | keystore: conf/.keystore |
| 794 | keystore_password: cassandra |
| 795 | truststore: conf/.truststore |
| 796 | truststore_password: cassandra |
| 797 | # More advanced defaults below: |
| 798 | # protocol: TLS |
| 799 | # algorithm: SunX509 |
| 800 | # store_type: JKS |
| 801 | # cipher_suites: [TLS_RSA_WITH_AES_128_CBC_SHA,TLS_RSA_WITH_AES_256_CBC_SHA,TLS_DHE_RSA_WITH_AES_128_CBC_SHA,TLS_DHE_RSA_WITH_AES_256_CBC_SHA,TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA,TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA] |
| 802 | # require_client_auth: false |
| 803 | |
| 804 | # enable or disable client/server encryption. |
| 805 | client_encryption_options: |
| 806 | enabled: false |
| 807 | # If enabled and optional is set to true encrypted and unencrypted connections are handled. |
| 808 | optional: false |
| 809 | keystore: conf/.keystore |
| 810 | keystore_password: cassandra |
| 811 | # require_client_auth: false |
| 812 | # Set trustore and truststore_password if require_client_auth is true |
| 813 | # truststore: conf/.truststore |
| 814 | # truststore_password: cassandra |
| 815 | # More advanced defaults below: |
| 816 | # protocol: TLS |
| 817 | # algorithm: SunX509 |
| 818 | # store_type: JKS |
| 819 | # cipher_suites: [TLS_RSA_WITH_AES_128_CBC_SHA,TLS_RSA_WITH_AES_256_CBC_SHA,TLS_DHE_RSA_WITH_AES_128_CBC_SHA,TLS_DHE_RSA_WITH_AES_256_CBC_SHA,TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA,TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA] |
| 820 | |
| 821 | # internode_compression controls whether traffic between nodes is |
| 822 | # compressed. |
| 823 | # can be: all - all traffic is compressed |
| 824 | # dc - traffic between different datacenters is compressed |
| 825 | # none - nothing is compressed. |
| 826 | internode_compression: all |
| 827 | |
| 828 | # Enable or disable tcp_nodelay for inter-dc communication. |
| 829 | # Disabling it will result in larger (but fewer) network packets being sent, |
| 830 | # reducing overhead from the TCP protocol itself, at the cost of increasing |
| 831 | # latency if you block for cross-datacenter responses. |
| 832 | inter_dc_tcp_nodelay: false |
| 833 | |
| 834 | # GC Pauses greater than gc_warn_threshold_in_ms will be logged at WARN level |
| 835 | # Adjust the threshold based on your application throughput requirement |
| 836 | # By default, Cassandra logs GC Pauses greater than 200 ms at INFO level |
| 837 | # gc_warn_threshold_in_ms: 1000 |