NFd6K7bg

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.
10cluster_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 on the node's initial start,
21# on subsequent starts, this setting will apply even if initial token is set.
22#
23# If you already have a cluster with 1 token per node, and wish to migrate to 
24# multiple tokens per node, see http://wiki.apache.org/cassandra/Operations
25num_tokens: 256
26
27# Triggers automatic allocation of num_tokens tokens for this node. The allocation
28# algorithm attempts to choose tokens in a way that optimizes replicated load over
29# the nodes in the datacenter for the replication strategy used by the specified
30# keyspace.
31#
32# The load assigned to each node will be close to proportional to its number of
33# vnodes.
34#
35# Only supported with the Murmur3Partitioner.
36# allocate_tokens_for_keyspace: KEYSPACE
37
38# initial_token allows you to specify tokens manually.  While you can use it with
39# vnodes (num_tokens > 1, above) -- in which case you should provide a 
40# comma-separated list -- it's primarily used when adding nodes to legacy clusters 
41# that do not have vnodes enabled.
42# initial_token:
43
44# See http://wiki.apache.org/cassandra/HintedHandoff
45# May either be "true" or "false" to enable globally
46hinted_handoff_enabled: true
47
48# When hinted_handoff_enabled is true, a black list of data centers that will not
49# perform hinted handoff
50# hinted_handoff_disabled_datacenters:
51#    - DC1
52#    - DC2
53
54# this defines the maximum amount of time a dead host will have hints
55# generated.  After it has been dead this long, new hints for it will not be
56# created until it has been seen alive and gone down again.
57max_hint_window_in_ms: 10800000 # 3 hours
58
59# Maximum throttle in KBs per second, per delivery thread.  This will be
60# reduced proportionally to the number of nodes in the cluster.  (If there
61# are two nodes in the cluster, each delivery thread will use the maximum
62# rate; if there are three, each will throttle to half of the maximum,
63# since we expect two nodes to be delivering hints simultaneously.)
64hinted_handoff_throttle_in_kb: 1024
65
66# Number of threads with which to deliver hints;
67# Consider increasing this number when you have multi-dc deployments, since
68# cross-dc handoff tends to be slower
69max_hints_delivery_threads: 2
70
71# Directory where Cassandra should store hints.
72# If not set, the default directory is $CASSANDRA_HOME/data/hints.
73# hints_directory: /var/lib/cassandra/hints
74
75# How often hints should be flushed from the internal buffers to disk.
76# Will *not* trigger fsync.
77hints_flush_period_in_ms: 10000
78
79# Maximum size for a single hints file, in megabytes.
80max_hints_file_size_in_mb: 128
81
82# Compression to apply to the hint files. If omitted, hints files
83# will be written uncompressed. LZ4, Snappy, and Deflate compressors
84# are supported.
85#hints_compression:
86#   - class_name: LZ4Compressor
87#     parameters:
88#         -
89
90# Maximum throttle in KBs per second, total. This will be
91# reduced proportionally to the number of nodes in the cluster.
92batchlog_replay_throttle_in_kb: 1024
93
94# Authentication backend, implementing IAuthenticator; used to identify users
95# Out of the box, Cassandra provides org.apache.cassandra.auth.{AllowAllAuthenticator,
96# PasswordAuthenticator}.
97#
98# - AllowAllAuthenticator performs no checks - set it to disable authentication.
99# - PasswordAuthenticator relies on username/password pairs to authenticate
100#   users. It keeps usernames and hashed passwords in system_auth.roles table.
101#   Please increase system_auth keyspace replication factor if you use this authenticator.
102#   If using PasswordAuthenticator, CassandraRoleManager must also be used (see below)
103authenticator: AllowAllAuthenticator
104
105# Authorization backend, implementing IAuthorizer; used to limit access/provide permissions
106# Out of the box, Cassandra provides org.apache.cassandra.auth.{AllowAllAuthorizer,
107# CassandraAuthorizer}.
108#
109# - AllowAllAuthorizer allows any action to any user - set it to disable authorization.
110# - CassandraAuthorizer stores permissions in system_auth.role_permissions table. Please
111#   increase system_auth keyspace replication factor if you use this authorizer.
112authorizer: AllowAllAuthorizer
113
114# Part of the Authentication & Authorization backend, implementing IRoleManager; used
115# to maintain grants and memberships between roles.
116# Out of the box, Cassandra provides org.apache.cassandra.auth.CassandraRoleManager,
117# which stores role information in the system_auth keyspace. Most functions of the
118# IRoleManager require an authenticated login, so unless the configured IAuthenticator
119# actually implements authentication, most of this functionality will be unavailable.
120#
121# - CassandraRoleManager stores role data in the system_auth keyspace. Please
122#   increase system_auth keyspace replication factor if you use this role manager.
123role_manager: CassandraRoleManager
124
125# Validity period for roles cache (fetching granted roles can be an expensive
126# operation depending on the role manager, CassandraRoleManager is one example)
127# Granted roles are cached for authenticated sessions in AuthenticatedUser and
128# after the period specified here, become eligible for (async) reload.
129# Defaults to 2000, set to 0 to disable caching entirely.
130# Will be disabled automatically for AllowAllAuthenticator.
131roles_validity_in_ms: 2000
132
133# Refresh interval for roles cache (if enabled).
134# After this interval, cache entries become eligible for refresh. Upon next
135# access, an async reload is scheduled and the old value returned until it
136# completes. If roles_validity_in_ms is non-zero, then this must be
137# also.
138# Defaults to the same value as roles_validity_in_ms.
139# roles_update_interval_in_ms: 2000
140
141# Validity period for permissions cache (fetching permissions can be an
142# expensive operation depending on the authorizer, CassandraAuthorizer is
143# one example). Defaults to 2000, set to 0 to disable.
144# Will be disabled automatically for AllowAllAuthorizer.
145permissions_validity_in_ms: 2000
146
147# Refresh interval for permissions cache (if enabled).
148# After this interval, cache entries become eligible for refresh. Upon next
149# access, an async reload is scheduled and the old value returned until it
150# completes. If permissions_validity_in_ms is non-zero, then this must be
151# also.
152# Defaults to the same value as permissions_validity_in_ms.
153# permissions_update_interval_in_ms: 2000
154
155# Validity period for credentials cache. This cache is tightly coupled to
156# the provided PasswordAuthenticator implementation of IAuthenticator. If
157# another IAuthenticator implementation is configured, this cache will not
158# be automatically used and so the following settings will have no effect.
159# Please note, credentials are cached in their encrypted form, so while
160# activating this cache may reduce the number of queries made to the
161# underlying table, it may not  bring a significant reduction in the
162# latency of individual authentication attempts.
163# Defaults to 2000, set to 0 to disable credentials caching.
164credentials_validity_in_ms: 2000
165
166# Refresh interval for credentials cache (if enabled).
167# After this interval, cache entries become eligible for refresh. Upon next
168# access, an async reload is scheduled and the old value returned until it
169# completes. If credentials_validity_in_ms is non-zero, then this must be
170# also.
171# Defaults to the same value as credentials_validity_in_ms.
172# credentials_update_interval_in_ms: 2000
173
174# The partitioner is responsible for distributing groups of rows (by
175# partition key) across nodes in the cluster.  You should leave this
176# alone for new clusters.  The partitioner can NOT be changed without
177# reloading all data, so when upgrading you should set this to the
178# same partitioner you were already using.
179#
180# Besides Murmur3Partitioner, partitioners included for backwards
181# compatibility include RandomPartitioner, ByteOrderedPartitioner, and
182# OrderPreservingPartitioner.
183#
184partitioner: org.apache.cassandra.dht.Murmur3Partitioner
185
186# Directories where Cassandra should store data on disk.  Cassandra
187# will spread data evenly across them, subject to the granularity of
188# the configured compaction strategy.
189# If not set, the default directory is $CASSANDRA_HOME/data/data.
190data_file_directories:
191    - /var/lib/cassandra/data
192
193# commit log.  when running on magnetic HDD, this should be a
194# separate spindle than the data directories.
195# If not set, the default directory is $CASSANDRA_HOME/data/commitlog.
196commitlog_directory: /var/lib/cassandra/commitlog
197
198# Enable / disable CDC functionality on a per-node basis. This modifies the logic used
199# for write path allocation rejection (standard: never reject. cdc: reject Mutation
200# containing a CDC-enabled table if at space limit in cdc_raw_directory).
201cdc_enabled: false
202
203# CommitLogSegments are moved to this directory on flush if cdc_enabled: true and the
204# segment contains mutations for a CDC-enabled table. This should be placed on a
205# separate spindle than the data directories. If not set, the default directory is
206# $CASSANDRA_HOME/data/cdc_raw.
207# cdc_raw_directory: /var/lib/cassandra/cdc_raw
208
209# Policy for data disk failures:
210#
211# die
212#   shut down gossip and client transports and kill the JVM for any fs errors or
213#   single-sstable errors, so the node can be replaced.
214#
215# stop_paranoid
216#   shut down gossip and client transports even for single-sstable errors,
217#   kill the JVM for errors during startup.
218#
219# stop
220#   shut down gossip and client transports, leaving the node effectively dead, but
221#   can still be inspected via JMX, kill the JVM for errors during startup.
222#
223# best_effort
224#    stop using the failed disk and respond to requests based on
225#    remaining available sstables.  This means you WILL see obsolete
226#    data at CL.ONE!
227#
228# ignore
229#    ignore fatal errors and let requests fail, as in pre-1.2 Cassandra
230disk_failure_policy: stop
231
232# Policy for commit disk failures:
233#
234# die
235#   shut down gossip and Thrift and kill the JVM, so the node can be replaced.
236#
237# stop
238#   shut down gossip and Thrift, leaving the node effectively dead, but
239#   can still be inspected via JMX.
240#
241# stop_commit
242#   shutdown the commit log, letting writes collect but
243#   continuing to service reads, as in pre-2.0.5 Cassandra
244#
245# ignore
246#   ignore fatal errors and let the batches fail
247commit_failure_policy: stop
248
249# Maximum size of the native protocol prepared statement cache
250#
251# Valid values are either "auto" (omitting the value) or a value greater 0.
252#
253# Note that specifying a too large value will result in long running GCs and possbily
254# out-of-memory errors. Keep the value at a small fraction of the heap.
255#
256# If you constantly see "prepared statements discarded in the last minute because
257# cache limit reached" messages, the first step is to investigate the root cause
258# of these messages and check whether prepared statements are used correctly -
259# i.e. use bind markers for variable parts.
260#
261# Do only change the default value, if you really have more prepared statements than
262# fit in the cache. In most cases it is not neccessary to change this value.
263# Constantly re-preparing statements is a performance penalty.
264#
265# Default value ("auto") is 1/256th of the heap or 10MB, whichever is greater
266prepared_statements_cache_size_mb:
267
268# Maximum size of the Thrift prepared statement cache
269#
270# If you do not use Thrift at all, it is safe to leave this value at "auto".
271#
272# See description of 'prepared_statements_cache_size_mb' above for more information.
273#
274# Default value ("auto") is 1/256th of the heap or 10MB, whichever is greater
275thrift_prepared_statements_cache_size_mb:
276
277# Maximum size of the key cache in memory.
278#
279# Each key cache hit saves 1 seek and each row cache hit saves 2 seeks at the
280# minimum, sometimes more. The key cache is fairly tiny for the amount of
281# time it saves, so it's worthwhile to use it at large numbers.
282# The row cache saves even more time, but must contain the entire row,
283# so it is extremely space-intensive. It's best to only use the
284# row cache if you have hot rows or static rows.
285#
286# NOTE: if you reduce the size, you may not get you hottest keys loaded on startup.
287#
288# Default value is empty to make it "auto" (min(5% of Heap (in MB), 100MB)). Set to 0 to disable key cache.
289key_cache_size_in_mb:
290
291# Duration in seconds after which Cassandra should
292# save the key cache. Caches are saved to saved_caches_directory as
293# specified in this configuration file.
294#
295# Saved caches greatly improve cold-start speeds, and is relatively cheap in
296# terms of I/O for the key cache. Row cache saving is much more expensive and
297# has limited use.
298#
299# Default is 14400 or 4 hours.
300key_cache_save_period: 14400
301
302# Number of keys from the key cache to save
303# Disabled by default, meaning all keys are going to be saved
304# key_cache_keys_to_save: 100
305
306# Row cache implementation class name. Available implementations:
307#
308# org.apache.cassandra.cache.OHCProvider
309#   Fully off-heap row cache implementation (default).
310#
311# org.apache.cassandra.cache.SerializingCacheProvider
312#   This is the row cache implementation availabile
313#   in previous releases of Cassandra.
314# row_cache_class_name: org.apache.cassandra.cache.OHCProvider
315
316# Maximum size of the row cache in memory.
317# Please note that OHC cache implementation requires some additional off-heap memory to manage
318# the map structures and some in-flight memory during operations before/after cache entries can be
319# accounted against the cache capacity. This overhead is usually small compared to the whole capacity.
320# Do not specify more memory that the system can afford in the worst usual situation and leave some
321# headroom for OS block level cache. Do never allow your system to swap.
322#
323# Default value is 0, to disable row caching.
324row_cache_size_in_mb: 0
325
326# Duration in seconds after which Cassandra should save the row cache.
327# Caches are saved to saved_caches_directory as specified in this configuration file.
328#
329# Saved caches greatly improve cold-start speeds, and is relatively cheap in
330# terms of I/O for the key cache. Row cache saving is much more expensive and
331# has limited use.
332#
333# Default is 0 to disable saving the row cache.
334row_cache_save_period: 0
335
336# Number of keys from the row cache to save.
337# Specify 0 (which is the default), meaning all keys are going to be saved
338# row_cache_keys_to_save: 100
339
340# Maximum size of the counter cache in memory.
341#
342# Counter cache helps to reduce counter locks' contention for hot counter cells.
343# In case of RF = 1 a counter cache hit will cause Cassandra to skip the read before
344# write entirely. With RF > 1 a counter cache hit will still help to reduce the duration
345# of the lock hold, helping with hot counter cell updates, but will not allow skipping
346# the read entirely. Only the local (clock, count) tuple of a counter cell is kept
347# in memory, not the whole counter, so it's relatively cheap.
348#
349# NOTE: if you reduce the size, you may not get you hottest keys loaded on startup.
350#
351# Default value is empty to make it "auto" (min(2.5% of Heap (in MB), 50MB)). Set to 0 to disable counter cache.
352# NOTE: if you perform counter deletes and rely on low gcgs, you should disable the counter cache.
353counter_cache_size_in_mb:
354
355# Duration in seconds after which Cassandra should
356# save the counter cache (keys only). Caches are saved to saved_caches_directory as
357# specified in this configuration file.
358#
359# Default is 7200 or 2 hours.
360counter_cache_save_period: 7200
361
362# Number of keys from the counter cache to save
363# Disabled by default, meaning all keys are going to be saved
364# counter_cache_keys_to_save: 100
365
366# saved caches
367# If not set, the default directory is $CASSANDRA_HOME/data/saved_caches.
368saved_caches_directory: /var/lib/cassandra/saved_caches
369
370# commitlog_sync may be either "periodic" or "batch." 
371# 
372# When in batch mode, Cassandra won't ack writes until the commit log
373# has been fsynced to disk.  It will wait
374# commitlog_sync_batch_window_in_ms milliseconds between fsyncs.
375# This window should be kept short because the writer threads will
376# be unable to do extra work while waiting.  (You may need to increase
377# concurrent_writes for the same reason.)
378#
379# commitlog_sync: batch
380# commitlog_sync_batch_window_in_ms: 2
381#
382# the other option is "periodic" where writes may be acked immediately
383# and the CommitLog is simply synced every commitlog_sync_period_in_ms
384# milliseconds.
385commitlog_sync: periodic
386commitlog_sync_period_in_ms: 10000
387
388# The size of the individual commitlog file segments.  A commitlog
389# segment may be archived, deleted, or recycled once all the data
390# in it (potentially from each columnfamily in the system) has been
391# flushed to sstables.
392#
393# The default size is 32, which is almost always fine, but if you are
394# archiving commitlog segments (see commitlog_archiving.properties),
395# then you probably want a finer granularity of archiving; 8 or 16 MB
396# is reasonable.
397# Max mutation size is also configurable via max_mutation_size_in_kb setting in
398# cassandra.yaml. The default is half the size commitlog_segment_size_in_mb * 1024.
399# This should be positive and less than 2048.
400#
401# NOTE: If max_mutation_size_in_kb is set explicitly then commitlog_segment_size_in_mb must
402# be set to at least twice the size of max_mutation_size_in_kb / 1024
403#
404commitlog_segment_size_in_mb: 32
405
406# Compression to apply to the commit log. If omitted, the commit log
407# will be written uncompressed.  LZ4, Snappy, and Deflate compressors
408# are supported.
409# commitlog_compression:
410#   - class_name: LZ4Compressor
411#     parameters:
412#         -
413
414# any class that implements the SeedProvider interface and has a
415# constructor that takes a Map<String, String> of parameters will do.
416seed_provider:
417    # Addresses of hosts that are deemed contact points. 
418    # Cassandra nodes use this list of hosts to find each other and learn
419    # the topology of the ring.  You must change this if you are running
420    # multiple nodes!
421    - class_name: org.apache.cassandra.locator.SimpleSeedProvider
422      parameters:
423          # seeds is actually a comma-delimited list of addresses.
424          # Ex: "<ip1>,<ip2>,<ip3>"
425          - seeds: "127.0.0.1"
426
427# For workloads with more data than can fit in memory, Cassandra's
428# bottleneck will be reads that need to fetch data from
429# disk. "concurrent_reads" should be set to (16 * number_of_drives) in
430# order to allow the operations to enqueue low enough in the stack
431# that the OS and drives can reorder them. Same applies to
432# "concurrent_counter_writes", since counter writes read the current
433# values before incrementing and writing them back.
434#
435# On the other hand, since writes are almost never IO bound, the ideal
436# number of "concurrent_writes" is dependent on the number of cores in
437# your system; (8 * number_of_cores) is a good rule of thumb.
438concurrent_reads: 10000
439concurrent_writes: 10000
440concurrent_counter_writes: 10000
441
442# For materialized view writes, as there is a read involved, so this should
443# be limited by the less of concurrent reads or concurrent writes.
444concurrent_materialized_view_writes: 9999
445
446# Maximum memory to use for sstable chunk cache and buffer pooling.
447# 32MB of this are reserved for pooling buffers, the rest is used as an
448# cache that holds uncompressed sstable chunks.
449# Defaults to the smaller of 1/4 of heap or 512MB. This pool is allocated off-heap,
450# so is in addition to the memory allocated for heap. The cache also has on-heap
451# overhead which is roughly 128 bytes per chunk (i.e. 0.2% of the reserved size
452# if the default 64k chunk size is used).
453# Memory is only allocated when needed.
454# file_cache_size_in_mb: 512
455
456# Flag indicating whether to allocate on or off heap when the sstable buffer
457# pool is exhausted, that is when it has exceeded the maximum memory
458# file_cache_size_in_mb, beyond which it will not cache buffers but allocate on request.
459
460# buffer_pool_use_heap_if_exhausted: true
461
462# The strategy for optimizing disk read
463# Possible values are:
464# ssd (for solid state disks, the default)
465# spinning (for spinning disks)
466# disk_optimization_strategy: ssd
467
468# Total permitted memory to use for memtables. Cassandra will stop
469# accepting writes when the limit is exceeded until a flush completes,
470# and will trigger a flush based on memtable_cleanup_threshold
471# If omitted, Cassandra will set both to 1/4 the size of the heap.
472# memtable_heap_space_in_mb: 2048
473# memtable_offheap_space_in_mb: 2048
474
475# memtable_cleanup_threshold is deprecated. The default calculation
476# is the only reasonable choice. See the comments on  memtable_flush_writers
477# for more information.
478#
479# Ratio of occupied non-flushing memtable size to total permitted size
480# that will trigger a flush of the largest memtable. Larger mct will
481# mean larger flushes and hence less compaction, but also less concurrent
482# flush activity which can make it difficult to keep your disks fed
483# under heavy write load.
484#
485# memtable_cleanup_threshold defaults to 1 / (memtable_flush_writers + 1)
486# memtable_cleanup_threshold: 0.11
487
488# Specify the way Cassandra allocates and manages memtable memory.
489# Options are:
490#
491# heap_buffers
492#   on heap nio buffers
493#
494# offheap_buffers
495#   off heap (direct) nio buffers
496#
497# offheap_objects
498#    off heap objects
499memtable_allocation_type: heap_buffers
500
501# Total space to use for commit logs on disk.
502#
503# If space gets above this value, Cassandra will flush every dirty CF
504# in the oldest segment and remove it.  So a small total commitlog space
505# will tend to cause more flush activity on less-active columnfamilies.
506#
507# The default value is the smaller of 8192, and 1/4 of the total space
508# of the commitlog volume.
509#
510# commitlog_total_space_in_mb: 8192
511
512# This sets the number of memtable flush writer threads per disk
513# as well as the total number of memtables that can be flushed concurrently.
514# These are generally a combination of compute and IO bound.
515#
516# Memtable flushing is more CPU efficient than memtable ingest and a single thread
517# can keep up with the ingest rate of a whole server on a single fast disk
518# until it temporarily becomes IO bound under contention typically with compaction.
519# At that point you need multiple flush threads. At some point in the future
520# it may become CPU bound all the time.
521#
522# You can tell if flushing is falling behind using the MemtablePool.BlockedOnAllocation
523# metric which should be 0, but will be non-zero if threads are blocked waiting on flushing
524# to free memory.
525#
526# memtable_flush_writers defaults to two for a single data directory.
527# This means that two  memtables can be flushed concurrently to the single data directory.
528# If you have multiple data directories the default is one memtable flushing at a time
529# but the flush will use a thread per data directory so you will get two or more writers.
530#
531# Two is generally enough to flush on a fast disk [array] mounted as a single data directory.
532# Adding more flush writers will result in smaller more frequent flushes that introduce more
533# compaction overhead.
534#
535# There is a direct tradeoff between number of memtables that can be flushed concurrently
536# and flush size and frequency. More is not better you just need enough flush writers
537# to never stall waiting for flushing to free memory.
538#
539#memtable_flush_writers: 2
540
541# Total space to use for change-data-capture logs on disk.
542#
543# If space gets above this value, Cassandra will throw WriteTimeoutException
544# on Mutations including tables with CDC enabled. A CDCCompactor is responsible
545# for parsing the raw CDC logs and deleting them when parsing is completed.
546#
547# The default value is the min of 4096 mb and 1/8th of the total space
548# of the drive where cdc_raw_directory resides.
549# cdc_total_space_in_mb: 4096
550
551# When we hit our cdc_raw limit and the CDCCompactor is either running behind
552# or experiencing backpressure, we check at the following interval to see if any
553# new space for cdc-tracked tables has been made available. Default to 250ms
554# cdc_free_space_check_interval_ms: 250
555
556# A fixed memory pool size in MB for for SSTable index summaries. If left
557# empty, this will default to 5% of the heap size. If the memory usage of
558# all index summaries exceeds this limit, SSTables with low read rates will
559# shrink their index summaries in order to meet this limit.  However, this
560# is a best-effort process. In extreme conditions Cassandra may need to use
561# more than this amount of memory.
562index_summary_capacity_in_mb:
563
564# How frequently index summaries should be resampled.  This is done
565# periodically to redistribute memory from the fixed-size pool to sstables
566# proportional their recent read rates.  Setting to -1 will disable this
567# process, leaving existing index summaries at their current sampling level.
568index_summary_resize_interval_in_minutes: 60
569
570# Whether to, when doing sequential writing, fsync() at intervals in
571# order to force the operating system to flush the dirty
572# buffers. Enable this to avoid sudden dirty buffer flushing from
573# impacting read latencies. Almost always a good idea on SSDs; not
574# necessarily on platters.
575trickle_fsync: false
576trickle_fsync_interval_in_kb: 10240
577
578# TCP port, for commands and data
579# For security reasons, you should not expose this port to the internet.  Firewall it if needed.
580storage_port: 7000
581
582# SSL port, for encrypted communication.  Unused unless enabled in
583# encryption_options
584# For security reasons, you should not expose this port to the internet.  Firewall it if needed.
585ssl_storage_port: 7001
586
587# Address or interface to bind to and tell other Cassandra nodes to connect to.
588# You _must_ change this if you want multiple nodes to be able to communicate!
589#
590# Set listen_address OR listen_interface, not both.
591#
592# Leaving it blank leaves it up to InetAddress.getLocalHost(). This
593# will always do the Right Thing _if_ the node is properly configured
594# (hostname, name resolution, etc), and the Right Thing is to use the
595# address associated with the hostname (it might not be).
596#
597# Setting listen_address to 0.0.0.0 is always wrong.
598#
599listen_address: localhost
600
601# Set listen_address OR listen_interface, not both. Interfaces must correspond
602# to a single address, IP aliasing is not supported.
603# listen_interface: eth0
604
605# If you choose to specify the interface by name and the interface has an ipv4 and an ipv6 address
606# you can specify which should be chosen using listen_interface_prefer_ipv6. If false the first ipv4
607# address will be used. If true the first ipv6 address will be used. Defaults to false preferring
608# ipv4. If there is only one address it will be selected regardless of ipv4/ipv6.
609# listen_interface_prefer_ipv6: false
610
611# Address to broadcast to other Cassandra nodes
612# Leaving this blank will set it to the same value as listen_address
613# broadcast_address: 1.2.3.4
614
615# When using multiple physical network interfaces, set this
616# to true to listen on broadcast_address in addition to
617# the listen_address, allowing nodes to communicate in both
618# interfaces.
619# Ignore this property if the network configuration automatically
620# routes  between the public and private networks such as EC2.
621# listen_on_broadcast_address: false
622
623# Internode authentication backend, implementing IInternodeAuthenticator;
624# used to allow/disallow connections from peer nodes.
625# internode_authenticator: org.apache.cassandra.auth.AllowAllInternodeAuthenticator
626
627# Whether to start the native transport server.
628# Please note that the address on which the native transport is bound is the
629# same as the rpc_address. The port however is different and specified below.
630start_native_transport: true
631# port for the CQL native transport to listen for clients on
632# For security reasons, you should not expose this port to the internet.  Firewall it if needed.
633native_transport_port: 9042
634# Enabling native transport encryption in client_encryption_options allows you to either use
635# encryption for the standard port or to use a dedicated, additional port along with the unencrypted
636# standard native_transport_port.
637# Enabling client encryption and keeping native_transport_port_ssl disabled will use encryption
638# for native_transport_port. Setting native_transport_port_ssl to a different value
639# from native_transport_port will use encryption for native_transport_port_ssl while
640# keeping native_transport_port unencrypted.
641# native_transport_port_ssl: 9142
642# The maximum threads for handling requests when the native transport is used.
643# This is similar to rpc_max_threads though the default differs slightly (and
644# there is no native_transport_min_threads, idle threads will always be stopped
645# after 30 seconds).
646# native_transport_max_threads: 128
647#
648# The maximum size of allowed frame. Frame (requests) larger than this will
649# be rejected as invalid. The default is 256MB. If you're changing this parameter,
650# you may want to adjust max_value_size_in_mb accordingly. This should be positive and less than 2048.
651# native_transport_max_frame_size_in_mb: 256
652
653# The maximum number of concurrent client connections.
654# The default is -1, which means unlimited.
655# native_transport_max_concurrent_connections: -1
656
657# The maximum number of concurrent client connections per source ip.
658# The default is -1, which means unlimited.
659# native_transport_max_concurrent_connections_per_ip: -1
660
661# Whether to start the thrift rpc server.
662start_rpc: false
663
664# The address or interface to bind the Thrift RPC service and native transport
665# server to.
666#
667# Set rpc_address OR rpc_interface, not both.
668#
669# Leaving rpc_address blank has the same effect as on listen_address
670# (i.e. it will be based on the configured hostname of the node).
671#
672# Note that unlike listen_address, you can specify 0.0.0.0, but you must also
673# set broadcast_rpc_address to a value other than 0.0.0.0.
674#
675# For security reasons, you should not expose this port to the internet.  Firewall it if needed.
676rpc_address: localhost
677
678# Set rpc_address OR rpc_interface, not both. Interfaces must correspond
679# to a single address, IP aliasing is not supported.
680# rpc_interface: eth1
681
682# If you choose to specify the interface by name and the interface has an ipv4 and an ipv6 address
683# you can specify which should be chosen using rpc_interface_prefer_ipv6. If false the first ipv4
684# address will be used. If true the first ipv6 address will be used. Defaults to false preferring
685# ipv4. If there is only one address it will be selected regardless of ipv4/ipv6.
686# rpc_interface_prefer_ipv6: false
687
688# port for Thrift to listen for clients on
689rpc_port: 9160
690
691# RPC address to broadcast to drivers and other Cassandra nodes. This cannot
692# be set to 0.0.0.0. If left blank, this will be set to the value of
693# rpc_address. If rpc_address is set to 0.0.0.0, broadcast_rpc_address must
694# be set.
695# broadcast_rpc_address: 1.2.3.4
696
697# enable or disable keepalive on rpc/native connections
698rpc_keepalive: true
699
700# Cassandra provides two out-of-the-box options for the RPC Server:
701#
702# sync
703#   One thread per thrift connection. For a very large number of clients, memory
704#   will be your limiting factor. On a 64 bit JVM, 180KB is the minimum stack size
705#   per thread, and that will correspond to your use of virtual memory (but physical memory
706#   may be limited depending on use of stack space).
707#
708# hsha
709#   Stands for "half synchronous, half asynchronous." All thrift clients are handled
710#   asynchronously using a small number of threads that does not vary with the amount
711#   of thrift clients (and thus scales well to many clients). The rpc requests are still
712#   synchronous (one thread per active request). If hsha is selected then it is essential
713#   that rpc_max_threads is changed from the default value of unlimited.
714#
715# The default is sync because on Windows hsha is about 30% slower.  On Linux,
716# sync/hsha performance is about the same, with hsha of course using less memory.
717#
718# Alternatively,  can provide your own RPC server by providing the fully-qualified class name
719# of an o.a.c.t.TServerFactory that can create an instance of it.
720rpc_server_type: sync
721
722# Uncomment rpc_min|max_thread to set request pool size limits.
723#
724# Regardless of your choice of RPC server (see above), the number of maximum requests in the
725# RPC thread pool dictates how many concurrent requests are possible (but if you are using the sync
726# RPC server, it also dictates the number of clients that can be connected at all).
727#
728# The default is unlimited and thus provides no protection against clients overwhelming the server. You are
729# encouraged to set a maximum that makes sense for you in production, but do keep in mind that
730# rpc_max_threads represents the maximum number of client requests this server may execute concurrently.
731#
732# rpc_min_threads: 16
733# rpc_max_threads: 2048
734
735# uncomment to set socket buffer sizes on rpc connections
736# rpc_send_buff_size_in_bytes:
737# rpc_recv_buff_size_in_bytes:
738
739# Uncomment to set socket buffer size for internode communication
740# Note that when setting this, the buffer size is limited by net.core.wmem_max
741# and when not setting it it is defined by net.ipv4.tcp_wmem
742# See also:
743# /proc/sys/net/core/wmem_max
744# /proc/sys/net/core/rmem_max
745# /proc/sys/net/ipv4/tcp_wmem
746# /proc/sys/net/ipv4/tcp_wmem
747# and 'man tcp'
748# internode_send_buff_size_in_bytes:
749
750# Uncomment to set socket buffer size for internode communication
751# Note that when setting this, the buffer size is limited by net.core.wmem_max
752# and when not setting it it is defined by net.ipv4.tcp_wmem
753# internode_recv_buff_size_in_bytes:
754
755# Frame size for thrift (maximum message length).
756thrift_framed_transport_size_in_mb: 15
757
758# Set to true to have Cassandra create a hard link to each sstable
759# flushed or streamed locally in a backups/ subdirectory of the
760# keyspace data.  Removing these links is the operator's
761# responsibility.
762incremental_backups: false
763
764# Whether or not to take a snapshot before each compaction.  Be
765# careful using this option, since Cassandra won't clean up the
766# snapshots for you.  Mostly useful if you're paranoid when there
767# is a data format change.
768snapshot_before_compaction: false
769
770# Whether or not a snapshot is taken of the data before keyspace truncation
771# or dropping of column families. The STRONGLY advised default of true 
772# should be used to provide data safety. If you set this flag to false, you will
773# lose data on truncation or drop.
774auto_snapshot: true
775
776# Granularity of the collation index of rows within a partition.
777# Increase if your rows are large, or if you have a very large
778# number of rows per partition.  The competing goals are these:
779#
780# - a smaller granularity means more index entries are generated
781#   and looking up rows withing the partition by collation column
782#   is faster
783# - but, Cassandra will keep the collation index in memory for hot
784#   rows (as part of the key cache), so a larger granularity means
785#   you can cache more hot rows
786column_index_size_in_kb: 64
787
788# Per sstable indexed key cache entries (the collation index in memory
789# mentioned above) exceeding this size will not be held on heap.
790# This means that only partition information is held on heap and the
791# index entries are read from disk.
792#
793# Note that this size refers to the size of the
794# serialized index information and not the size of the partition.
795column_index_cache_size_in_kb: 2
796
797# Number of simultaneous compactions to allow, NOT including
798# validation "compactions" for anti-entropy repair.  Simultaneous
799# compactions can help preserve read performance in a mixed read/write
800# workload, by mitigating the tendency of small sstables to accumulate
801# during a single long running compactions. The default is usually
802# fine and if you experience problems with compaction running too
803# slowly or too fast, you should look at
804# compaction_throughput_mb_per_sec first.
805#
806# concurrent_compactors defaults to the smaller of (number of disks,
807# number of cores), with a minimum of 2 and a maximum of 8.
808# 
809# If your data directories are backed by SSD, you should increase this
810# to the number of cores.
811#concurrent_compactors: 1
812
813# Throttles compaction to the given total throughput across the entire
814# system. The faster you insert data, the faster you need to compact in
815# order to keep the sstable count down, but in general, setting this to
816# 16 to 32 times the rate you are inserting data is more than sufficient.
817# Setting this to 0 disables throttling. Note that this account for all types
818# of compaction, including validation compaction.
819compaction_throughput_mb_per_sec: 16
820
821# When compacting, the replacement sstable(s) can be opened before they
822# are completely written, and used in place of the prior sstables for
823# any range that has been written. This helps to smoothly transfer reads 
824# between the sstables, reducing page cache churn and keeping hot rows hot
825sstable_preemptive_open_interval_in_mb: 50
826
827# Throttles all outbound streaming file transfers on this node to the
828# given total throughput in Mbps. This is necessary because Cassandra does
829# mostly sequential IO when streaming data during bootstrap or repair, which
830# can lead to saturating the network connection and degrading rpc performance.
831# When unset, the default is 200 Mbps or 25 MB/s.
832# stream_throughput_outbound_megabits_per_sec: 200
833
834# Throttles all streaming file transfer between the datacenters,
835# this setting allows users to throttle inter dc stream throughput in addition
836# to throttling all network stream traffic as configured with
837# stream_throughput_outbound_megabits_per_sec
838# When unset, the default is 200 Mbps or 25 MB/s
839# inter_dc_stream_throughput_outbound_megabits_per_sec: 200
840
841# How long the coordinator should wait for read operations to complete
842read_request_timeout_in_ms: 5000
843# How long the coordinator should wait for seq or index scans to complete
844range_request_timeout_in_ms: 10000
845# How long the coordinator should wait for writes to complete
846write_request_timeout_in_ms: 2000
847# How long the coordinator should wait for counter writes to complete
848counter_write_request_timeout_in_ms: 5000
849# How long a coordinator should continue to retry a CAS operation
850# that contends with other proposals for the same row
851cas_contention_timeout_in_ms: 1000
852# How long the coordinator should wait for truncates to complete
853# (This can be much longer, because unless auto_snapshot is disabled
854# we need to flush first so we can snapshot before removing the data.)
855truncate_request_timeout_in_ms: 60000
856# The default timeout for other, miscellaneous operations
857request_timeout_in_ms: 10000
858
859# How long before a node logs slow queries. Select queries that take longer than
860# this timeout to execute, will generate an aggregated log message, so that slow queries
861# can be identified. Set this value to zero to disable slow query logging.
862slow_query_log_timeout_in_ms: 500
863
864# Enable operation timeout information exchange between nodes to accurately
865# measure request timeouts.  If disabled, replicas will assume that requests
866# were forwarded to them instantly by the coordinator, which means that
867# under overload conditions we will waste that much extra time processing 
868# already-timed-out requests.
869#
870# Warning: before enabling this property make sure to ntp is installed
871# and the times are synchronized between the nodes.
872cross_node_timeout: false
873
874# Set keep-alive period for streaming
875# This node will send a keep-alive message periodically with this period.
876# If the node does not receive a keep-alive message from the peer for
877# 2 keep-alive cycles the stream session times out and fail
878# Default value is 300s (5 minutes), which means stalled stream
879# times out in 10 minutes by default
880# streaming_keep_alive_period_in_secs: 300
881
882# phi value that must be reached for a host to be marked down.
883# most users should never need to adjust this.
884# phi_convict_threshold: 8
885
886# endpoint_snitch -- Set this to a class that implements
887# IEndpointSnitch.  The snitch has two functions:
888#
889# - it teaches Cassandra enough about your network topology to route
890#   requests efficiently
891# - it allows Cassandra to spread replicas around your cluster to avoid
892#   correlated failures. It does this by grouping machines into
893#   "datacenters" and "racks."  Cassandra will do its best not to have
894#   more than one replica on the same "rack" (which may not actually
895#   be a physical location)
896#
897# CASSANDRA WILL NOT ALLOW YOU TO SWITCH TO AN INCOMPATIBLE SNITCH
898# ONCE DATA IS INSERTED INTO THE CLUSTER.  This would cause data loss.
899# This means that if you start with the default SimpleSnitch, which
900# locates every node on "rack1" in "datacenter1", your only options
901# if you need to add another datacenter are GossipingPropertyFileSnitch
902# (and the older PFS).  From there, if you want to migrate to an
903# incompatible snitch like Ec2Snitch you can do it by adding new nodes
904# under Ec2Snitch (which will locate them in a new "datacenter") and
905# decommissioning the old ones.
906#
907# Out of the box, Cassandra provides:
908#
909# SimpleSnitch:
910#    Treats Strategy order as proximity. This can improve cache
911#    locality when disabling read repair.  Only appropriate for
912#    single-datacenter deployments.
913#
914# GossipingPropertyFileSnitch
915#    This should be your go-to snitch for production use.  The rack
916#    and datacenter for the local node are defined in
917#    cassandra-rackdc.properties and propagated to other nodes via
918#    gossip.  If cassandra-topology.properties exists, it is used as a
919#    fallback, allowing migration from the PropertyFileSnitch.
920#
921# PropertyFileSnitch:
922#    Proximity is determined by rack and data center, which are
923#    explicitly configured in cassandra-topology.properties.
924#
925# Ec2Snitch:
926#    Appropriate for EC2 deployments in a single Region. Loads Region
927#    and Availability Zone information from the EC2 API. The Region is
928#    treated as the datacenter, and the Availability Zone as the rack.
929#    Only private IPs are used, so this will not work across multiple
930#    Regions.
931#
932# Ec2MultiRegionSnitch:
933#    Uses public IPs as broadcast_address to allow cross-region
934#    connectivity.  (Thus, you should set seed addresses to the public
935#    IP as well.) You will need to open the storage_port or
936#    ssl_storage_port on the public IP firewall.  (For intra-Region
937#    traffic, Cassandra will switch to the private IP after
938#    establishing a connection.)
939#
940# RackInferringSnitch:
941#    Proximity is determined by rack and data center, which are
942#    assumed to correspond to the 3rd and 2nd octet of each node's IP
943#    address, respectively.  Unless this happens to match your
944#    deployment conventions, this is best used as an example of
945#    writing a custom Snitch class and is provided in that spirit.
946#
947# You can use a custom Snitch by setting this to the full class name
948# of the snitch, which will be assumed to be on your classpath.
949endpoint_snitch: SimpleSnitch
950
951# controls how often to perform the more expensive part of host score
952# calculation
953dynamic_snitch_update_interval_in_ms: 100 
954# controls how often to reset all host scores, allowing a bad host to
955# possibly recover
956dynamic_snitch_reset_interval_in_ms: 600000
957# if set greater than zero and read_repair_chance is < 1.0, this will allow
958# 'pinning' of replicas to hosts in order to increase cache capacity.
959# The badness threshold will control how much worse the pinned host has to be
960# before the dynamic snitch will prefer other replicas over it.  This is
961# expressed as a double which represents a percentage.  Thus, a value of
962# 0.2 means Cassandra would continue to prefer the static snitch values
963# until the pinned host was 20% worse than the fastest.
964dynamic_snitch_badness_threshold: 0.1
965
966# request_scheduler -- Set this to a class that implements
967# RequestScheduler, which will schedule incoming client requests
968# according to the specific policy. This is useful for multi-tenancy
969# with a single Cassandra cluster.
970# NOTE: This is specifically for requests from the client and does
971# not affect inter node communication.
972# org.apache.cassandra.scheduler.NoScheduler - No scheduling takes place
973# org.apache.cassandra.scheduler.RoundRobinScheduler - Round robin of
974# client requests to a node with a separate queue for each
975# request_scheduler_id. The scheduler is further customized by
976# request_scheduler_options as described below.
977request_scheduler: org.apache.cassandra.scheduler.NoScheduler
978
979# Scheduler Options vary based on the type of scheduler
980#
981# NoScheduler
982#   Has no options
983#
984# RoundRobin
985#   throttle_limit
986#     The throttle_limit is the number of in-flight
987#     requests per client.  Requests beyond 
988#     that limit are queued up until
989#     running requests can complete.
990#     The value of 80 here is twice the number of
991#     concurrent_reads + concurrent_writes.
992#   default_weight
993#     default_weight is optional and allows for
994#     overriding the default which is 1.
995#   weights
996#     Weights are optional and will default to 1 or the
997#     overridden default_weight. The weight translates into how
998#     many requests are handled during each turn of the
999#     RoundRobin, based on the scheduler id.
1000#
1001# request_scheduler_options:
1002#    throttle_limit: 80
1003#    default_weight: 5
1004#    weights:
1005#      Keyspace1: 1
1006#      Keyspace2: 5
1007
1008# request_scheduler_id -- An identifier based on which to perform
1009# the request scheduling. Currently the only valid option is keyspace.
1010# request_scheduler_id: keyspace
1011
1012# Enable or disable inter-node encryption
1013# JVM defaults for supported SSL socket protocols and cipher suites can
1014# be replaced using custom encryption options. This is not recommended
1015# unless you have policies in place that dictate certain settings, or
1016# need to disable vulnerable ciphers or protocols in case the JVM cannot
1017# be updated.
1018# FIPS compliant settings can be configured at JVM level and should not
1019# involve changing encryption settings here:
1020# https://docs.oracle.com/javase/8/docs/technotes/guides/security/jsse/FIPS.html
1021# *NOTE* No custom encryption options are enabled at the moment
1022# The available internode options are : all, none, dc, rack
1023#
1024# If set to dc cassandra will encrypt the traffic between the DCs
1025# If set to rack cassandra will encrypt the traffic between the racks
1026#
1027# The passwords used in these options must match the passwords used when generating
1028# the keystore and truststore.  For instructions on generating these files, see:
1029# http://download.oracle.com/javase/6/docs/technotes/guides/security/jsse/JSSERefGuide.html#CreateKeystore
1030#
1031server_encryption_options:
1032    internode_encryption: none
1033    keystore: conf/.keystore
1034    keystore_password: cassandra
1035    truststore: conf/.truststore
1036    truststore_password: cassandra
1037    # More advanced defaults below:
1038    # protocol: TLS
1039    # algorithm: SunX509
1040    # store_type: JKS
1041    # 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]
1042    # require_client_auth: false
1043    # require_endpoint_verification: false
1044
1045# enable or disable client/server encryption.
1046client_encryption_options:
1047    enabled: false
1048    # If enabled and optional is set to true encrypted and unencrypted connections are handled.
1049    optional: false
1050    keystore: conf/.keystore
1051    keystore_password: cassandra
1052    # require_client_auth: false
1053    # Set trustore and truststore_password if require_client_auth is true
1054    # truststore: conf/.truststore
1055    # truststore_password: cassandra
1056    # More advanced defaults below:
1057    # protocol: TLS
1058    # algorithm: SunX509
1059    # store_type: JKS
1060    # 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]
1061
1062# internode_compression controls whether traffic between nodes is
1063# compressed.
1064# Can be:
1065#
1066# all
1067#   all traffic is compressed
1068#
1069# dc
1070#   traffic between different datacenters is compressed
1071#
1072# none
1073#   nothing is compressed.
1074internode_compression: dc
1075
1076# Enable or disable tcp_nodelay for inter-dc communication.
1077# Disabling it will result in larger (but fewer) network packets being sent,
1078# reducing overhead from the TCP protocol itself, at the cost of increasing
1079# latency if you block for cross-datacenter responses.
1080inter_dc_tcp_nodelay: false
1081
1082# TTL for different trace types used during logging of the repair process.
1083tracetype_query_ttl: 86400
1084tracetype_repair_ttl: 604800
1085
1086# By default, Cassandra logs GC Pauses greater than 200 ms at INFO level
1087# This threshold can be adjusted to minimize logging if necessary
1088# gc_log_threshold_in_ms: 200
1089
1090# If unset, all GC Pauses greater than gc_log_threshold_in_ms will log at
1091# INFO level
1092# UDFs (user defined functions) are disabled by default.
1093# As of Cassandra 3.0 there is a sandbox in place that should prevent execution of evil code.
1094enable_user_defined_functions: false
1095
1096# Enables scripted UDFs (JavaScript UDFs).
1097# Java UDFs are always enabled, if enable_user_defined_functions is true.
1098# Enable this option to be able to use UDFs with "language javascript" or any custom JSR-223 provider.
1099# This option has no effect, if enable_user_defined_functions is false.
1100enable_scripted_user_defined_functions: false
1101
1102# Enables materialized view creation on this node.
1103# Materialized views are considered experimental and are not recommended for production use.
1104enable_materialized_views: true
1105
1106# The default Windows kernel timer and scheduling resolution is 15.6ms for power conservation.
1107# Lowering this value on Windows can provide much tighter latency and better throughput, however
1108# some virtualized environments may see a negative performance impact from changing this setting
1109# below their system default. The sysinternals 'clockres' tool can confirm your system's default
1110# setting.
1111windows_timer_interval: 1
1112
1113
1114# Enables encrypting data at-rest (on disk). Different key providers can be plugged in, but the default reads from
1115# a JCE-style keystore. A single keystore can hold multiple keys, but the one referenced by
1116# the "key_alias" is the only key that will be used for encrypt opertaions; previously used keys
1117# can still (and should!) be in the keystore and will be used on decrypt operations
1118# (to handle the case of key rotation).
1119#
1120# It is strongly recommended to download and install Java Cryptography Extension (JCE)
1121# Unlimited Strength Jurisdiction Policy Files for your version of the JDK.
1122# (current link: http://www.oracle.com/technetwork/java/javase/downloads/jce8-download-2133166.html)
1123#
1124# Currently, only the following file types are supported for transparent data encryption, although
1125# more are coming in future cassandra releases: commitlog, hints
1126transparent_data_encryption_options:
1127    enabled: false
1128    chunk_length_kb: 64
1129    cipher: AES/CBC/PKCS5Padding
1130    key_alias: testing:1
1131    # CBC IV length for AES needs to be 16 bytes (which is also the default size)
1132    # iv_length: 16
1133    key_provider: 
1134      - class_name: org.apache.cassandra.security.JKSKeyProvider
1135        parameters: 
1136          - keystore: conf/.keystore
1137            keystore_password: cassandra
1138            store_type: JCEKS
1139            key_password: cassandra
1140
1141
1142#####################
1143# SAFETY THRESHOLDS #
1144#####################
1145
1146# When executing a scan, within or across a partition, we need to keep the
1147# tombstones seen in memory so we can return them to the coordinator, which
1148# will use them to make sure other replicas also know about the deleted rows.
1149# With workloads that generate a lot of tombstones, this can cause performance
1150# problems and even exaust the server heap.
1151# (http://www.datastax.com/dev/blog/cassandra-anti-patterns-queues-and-queue-like-datasets)
1152# Adjust the thresholds here if you understand the dangers and want to
1153# scan more tombstones anyway.  These thresholds may also be adjusted at runtime
1154# using the StorageService mbean.
1155tombstone_warn_threshold: 1000
1156tombstone_failure_threshold: 100000
1157
1158# Log WARN on any multiple-partition batch size exceeding this value. 5kb per batch by default.
1159# Caution should be taken on increasing the size of this threshold as it can lead to node instability.
1160batch_size_warn_threshold_in_kb: 5
1161
1162# Fail any multiple-partition batch exceeding this value. 50kb (10x warn threshold) by default.
1163batch_size_fail_threshold_in_kb: 50
1164
1165# Log WARN on any batches not of type LOGGED than span across more partitions than this limit
1166unlogged_batch_across_partitions_warn_threshold: 10
1167
1168# Log a warning when compacting partitions larger than this value
1169compaction_large_partition_warning_threshold_mb: 100
1170
1171# GC Pauses greater than gc_warn_threshold_in_ms will be logged at WARN level
1172# Adjust the threshold based on your application throughput requirement
1173# By default, Cassandra logs GC Pauses greater than 200 ms at INFO level
1174gc_warn_threshold_in_ms: 1000
1175
1176# Maximum size of any value in SSTables. Safety measure to detect SSTable corruption
1177# early. Any value size larger than this threshold will result into marking an SSTable
1178# as corrupted. This should be positive and less than 2048.
1179# max_value_size_in_mb: 256
1180
1181# Back-pressure settings #
1182# If enabled, the coordinator will apply the back-pressure strategy specified below to each mutation
1183# sent to replicas, with the aim of reducing pressure on overloaded replicas.
1184back_pressure_enabled: false
1185# The back-pressure strategy applied.
1186# The default implementation, RateBasedBackPressure, takes three arguments:
1187# high ratio, factor, and flow type, and uses the ratio between incoming mutation responses and outgoing mutation requests.
1188# If below high ratio, outgoing mutations are rate limited according to the incoming rate decreased by the given factor;
1189# if above high ratio, the rate limiting is increased by the given factor;
1190# such factor is usually best configured between 1 and 10, use larger values for a faster recovery
1191# at the expense of potentially more dropped mutations;
1192# the rate limiting is applied according to the flow type: if FAST, it's rate limited at the speed of the fastest replica,
1193# if SLOW at the speed of the slowest one.
1194# New strategies can be added. Implementors need to implement org.apache.cassandra.net.BackpressureStrategy and
1195# provide a public constructor accepting a Map<String, Object>.
1196back_pressure_strategy:
1197    - class_name: org.apache.cassandra.net.RateBasedBackPressure
1198      parameters:
1199        - high_ratio: 0.90
1200          factor: 5
1201          flow: FAST
1202
1203# Coalescing Strategies #
1204# Coalescing multiples messages turns out to significantly boost message processing throughput (think doubling or more).
1205# On bare metal, the floor for packet processing throughput is high enough that many applications won't notice, but in
1206# virtualized environments, the point at which an application can be bound by network packet processing can be
1207# surprisingly low compared to the throughput of task processing that is possible inside a VM. It's not that bare metal
1208# doesn't benefit from coalescing messages, it's that the number of packets a bare metal network interface can process
1209# is sufficient for many applications such that no load starvation is experienced even without coalescing.
1210# There are other benefits to coalescing network messages that are harder to isolate with a simple metric like messages
1211# per second. By coalescing multiple tasks together, a network thread can process multiple messages for the cost of one
1212# trip to read from a socket, and all the task submission work can be done at the same time reducing context switching
1213# and increasing cache friendliness of network message processing.
1214# See CASSANDRA-8692 for details.
1215
1216# Strategy to use for coalescing messages in OutboundTcpConnection.
1217# Can be fixed, movingaverage, timehorizon, disabled (default).
1218# You can also specify a subclass of CoalescingStrategies.CoalescingStrategy by name.
1219# otc_coalescing_strategy: DISABLED
1220
1221# How many microseconds to wait for coalescing. For fixed strategy this is the amount of time after the first
1222# message is received before it will be sent with any accompanying messages. For moving average this is the
1223# maximum amount of time that will be waited as well as the interval at which messages must arrive on average
1224# for coalescing to be enabled.
1225# otc_coalescing_window_us: 200
1226
1227# Do not try to coalesce messages if we already got that many messages. This should be more than 2 and less than 128.
1228# otc_coalescing_enough_coalesced_messages: 8
1229
1230# How many milliseconds to wait between two expiration runs on the backlog (queue) of the OutboundTcpConnection.
1231# Expiration is done if messages are piling up in the backlog. Droppable messages are expired to free the memory
1232# taken by expired messages. The interval should be between 0 and 1000, and in most installations the default value
1233# will be appropriate. A smaller value could potentially expire messages slightly sooner at the expense of more CPU
1234# time and queue contention while iterating the backlog of messages.
1235# An interval of 0 disables any wait time, which is the behavior of former Cassandra versions.
1236#
1237# otc_backlog_expiration_interval_ms: 200