[vbhang...@gmail.com]

-- Consistency level  LQ
-- It started happening approximately couple of months back.  Issue is very 
inconsistent and can't be reproduced.  It used rarely happen earlier (since 
last few years).
-- There are very few GC pauses but  they don't coincide with the issue. 
-- 99% latency is less than 80ms and 75% is less than 5ms.

- Vedant
On 2017-10-22 21:29, Jeff Jirsa <jji...@gmail.com> wrote: 
> What consistency level do you use on writes?
> Did this just start or has it always happened ?
> Are you seeing GC pauses at all?
> 
> What’s your 99% write latency? 
> 
> -- 
> Jeff Jirsa
> 
> 
> > On Oct 22, 2017, at 9:21 PM, "vbhang...@gmail.com"<vbhang...@gmail.com> 
> > wrote:
> > 
> > This is for Cassandra 2.1.13. At times there are replication delays across 
> > multiple regions. Data is available (getting queried from command line) in 
> > 1 region but not seen in other region(s).  This is not consistent. It is 
> > cluster spanning multiple data centers with total > 30 nodes. Keyspace is 
> > configured to get replicated in all the data centers.
> > 
> > Hints are getting piled up in the source region. This happens especially 
> > for large data payload (appro 1kb to few MB blobs).  Network  level 
> > congestion or saturation does not seem to be an issue.  There is no 
> > memory/cpu pressure on individual nodes.
> > 
> > I am sharing Cassandra.yaml below, any pointers on what can be tuned are 
> > highly appreciated. Let me know if you need any other info.
> > 
> > We tried bumping up hinted_handoff_throttle_in_kb: 30720 and handoff tends 
> > to be slower max_hints_delivery_threads: 12 on one of the nodes to see if 
> > it speeds up hints delivery, there was some improvement but not whole lot.
> > 
> > Thanks
> > 
> > =================================
> > # Cassandra storage config YAML
> > 
> > # NOTE:
> > #   See http://wiki.apache.org/cassandra/StorageConfiguration for
> > #   full explanations of configuration directives
> > # /NOTE
> > 
> > # The name of the cluster. This is mainly used to prevent machines in
> > # one logical cluster from joining another.
> > cluster_name: "central"
> > 
> > # This defines the number of tokens randomly assigned to this node on the 
> > ring
> > # The more tokens, relative to other nodes, the larger the proportion of 
> > data
> > # that this node will store. You probably want all nodes to have the same 
> > number
> > # of tokens assuming they have equal hardware capability.
> > #
> > # If you leave this unspecified, Cassandra will use the default of 1 token 
> > for legacy compatibility,
> > # and will use the initial_token as described below.
> > #
> > # Specifying initial_token will override this setting on the node's initial 
> > start,
> > # on subsequent starts, this setting will apply even if initial token is 
> > set.
> > #
> > # If you already have a cluster with 1 token per node, and wish to migrate 
> > to
> > # multiple tokens per node, see http://wiki.apache.org/cassandra/Operations
> > #num_tokens: 256
> > 
> > # initial_token allows you to specify tokens manually.  While you can use # 
> > it with
> > # vnodes (num_tokens > 1, above) -- in which case you should provide a
> > # comma-separated list -- it's primarily used when adding nodes # to legacy 
> > clusters
> > # that do not have vnodes enabled.
> > # initial_token:
> > 
> > initial_token: xxxx
> > 
> > # See http://wiki.apache.org/cassandra/HintedHandoff
> > # May either be "true" or "false" to enable globally, or contain a list
> > # of data centers to enable per-datacenter.
> > # hinted_handoff_enabled: DC1,DC2
> > hinted_handoff_enabled: true
> > # this defines the maximum amount of time a dead host will have hints
> > # generated.  After it has been dead this long, new hints for it will not be
> > # created until it has been seen alive and gone down again.
> > max_hint_window_in_ms: 10800000 # 3 hours
> > # Maximum throttle in KBs per second, per delivery thread.  This will be
> > # reduced proportionally to the number of nodes in the cluster.  (If there
> > # are two nodes in the cluster, each delivery thread will use the maximum
> > # rate; if there are three, each will throttle to half of the maximum,
> > # since we expect two nodes to be delivering hints simultaneously.)
> > hinted_handoff_throttle_in_kb: 1024
> > # Number of threads with which to deliver hints;
> > # Consider increasing this number when you have multi-dc deployments, since
> > # cross-dc handoff tends to be slower
> > max_hints_delivery_threads: 6
> > 
> > # Maximum throttle in KBs per second, total. This will be
> > # reduced proportionally to the number of nodes in the cluster.
> > batchlog_replay_throttle_in_kb: 1024
> > 
> > # Authentication backend, implementing IAuthenticator; used to identify 
> > users
> > # Out of the box, Cassandra provides 
> > org.apache.cassandra.auth.{AllowAllAuthenticator,
> > # PasswordAuthenticator}.
> > #
> > # - AllowAllAuthenticator performs no checks - set it to disable 
> > authentication.
> > # - PasswordAuthenticator relies on username/password pairs to authenticate
> > #   users. It keeps usernames and hashed passwords in 
> > system_auth.credentials table.
> > #   Please increase system_auth keyspace replication factor if you use this 
> > authenticator.
> > authenticator: AllowAllAuthenticator
> > 
> > # Authorization backend, implementing IAuthorizer; used to limit 
> > access/provide permissions
> > # Out of the box, Cassandra provides 
> > org.apache.cassandra.auth.{AllowAllAuthorizer,
> > # CassandraAuthorizer}.
> > #
> > # - AllowAllAuthorizer allows any action to any user - set it to disable 
> > authorization.
> > # - CassandraAuthorizer stores permissions in system_auth.permissions 
> > table. Please
> > #   increase system_auth keyspace replication factor if you use this 
> > authorizer.
> > authorizer: AllowAllAuthorizer
> > 
> > # Validity period for permissions cache (fetching permissions can be an
> > # expensive operation depending on the authorizer, CassandraAuthorizer is
> > # one example). Defaults to 2000, set to 0 to disable.
> > # Will be disabled automatically for AllowAllAuthorizer.
> > permissions_validity_in_ms: 2000
> > 
> > # Refresh interval for permissions cache (if enabled).
> > # After this interval, cache entries become eligible for refresh. Upon next
> > # access, an async reload is scheduled and the old value returned until it
> > # completes. If permissions_validity_in_ms is non-zero, then this must be
> > # also.
> > # Defaults to the same value as permissions_validity_in_ms.
> > # permissions_update_interval_in_ms: 1000
> > 
> > # The partitioner is responsible for distributing groups of rows (by
> > # partition key) across nodes in the cluster.  You should leave this
> > # alone for new clusters.  The partitioner can NOT be changed without
> > # reloading all data, so when upgrading you should set this to the
> > # same partitioner you were already using.
> > #
> > # Besides Murmur3Partitioner, partitioners included for backwards
> > # compatibility include RandomPartitioner, ByteOrderedPartitioner, and
> > # OrderPreservingPartitioner.
> > #
> > partitioner: org.apache.cassandra.dht.RandomPartitioner
> > 
> > # Directories where Cassandra should store data on disk.  Cassandra
> > # will spread data evenly across them, subject to the granularity of
> > # the configured compaction strategy.
> > # If not set, the default directory is $CASSANDRA_HOME/data/data.
> > data_file_directories:
> >     - /var/lib/cassandra/data
> > 
> > # commit log.  when running on magnetic HDD, this should be a
> > # separate spindle than the data directories.
> > # If not set, the default directory is $CASSANDRA_HOME/data/commitlog.
> > commitlog_directory: /data/cassandra/commitlog
> > 
> > # policy for data disk failures:
> > # die: shut down gossip and client transports and kill the JVM for any fs 
> > errors or
> > #      single-sstable errors, so the node can be replaced.
> > # stop_paranoid: shut down gossip and client transports even for 
> > single-sstable errors,
> > #                kill the JVM for errors during startup.
> > # stop: shut down gossip and client transports, leaving the node 
> > effectively dead, but
> > #       can still be inspected via JMX, kill the JVM for errors during 
> > startup.
> > # best_effort: stop using the failed disk and respond to requests based on
> > #              remaining available sstables.  This means you WILL see 
> > obsolete
> > #              data at CL.ONE!
> > # ignore: ignore fatal errors and let requests fail, as in pre-1.2 Cassandra
> > disk_failure_policy: stop
> > 
> > # policy for commit disk failures:
> > # die: shut down gossip and Thrift and kill the JVM, so the node can be 
> > replaced.
> > # stop: shut down gossip and Thrift, leaving the node effectively dead, but
> > #       can still be inspected via JMX.
> > # stop_commit: shutdown the commit log, letting writes collect but
> > #              continuing to service reads, as in pre-2.0.5 Cassandra
> > # ignore: ignore fatal errors and let the batches fail
> > commit_failure_policy: stop
> > 
> > # Maximum size of the key cache in memory.
> > #
> > # Each key cache hit saves 1 seek and each row cache hit saves 2 seeks at 
> > the
> > # minimum, sometimes more. The key cache is fairly tiny for the amount of
> > # time it saves, so it's worthwhile to use it at large numbers.
> > # The row cache saves even more time, but must contain the entire row,
> > # so it is extremely space-intensive. It's best to only use the
> > # row cache if you have hot rows or static rows.
> > #
> > # NOTE: if you reduce the size, you may not get you hottest keys loaded on 
> > startup.
> > #
> > # Default value is empty to make it "auto" (min(5% of Heap (in MB), 
> > 100MB)). Set to 0 to disable key cache.
> > key_cache_size_in_mb:
> > 
> > # Duration in seconds after which Cassandra should
> > # save the key cache. Caches are saved to saved_caches_directory as
> > # specified in this configuration file.
> > #
> > # Saved caches greatly improve cold-start speeds, and is relatively cheap in
> > # terms of I/O for the key cache. Row cache saving is much more expensive 
> > and
> > # has limited use.
> > #
> > # Default is 14400 or 4 hours.
> > key_cache_save_period: 14400
> > 
> > # Number of keys from the key cache to save
> > # Disabled by default, meaning all keys are going to be saved
> > # key_cache_keys_to_save: 100
> > 
> > # Maximum size of the row cache in memory.
> > # NOTE: if you reduce the size, you may not get you hottest keys loaded on 
> > startup.
> > #
> > # Default value is 0, to disable row caching.
> > row_cache_size_in_mb: 0
> > 
> > # Duration in seconds after which Cassandra should
> > # save the row cache. Caches are saved to saved_caches_directory as 
> > specified
> > # in this configuration file.
> > #
> > # Saved caches greatly improve cold-start speeds, and is relatively cheap in
> > # terms of I/O for the key cache. Row cache saving is much more expensive 
> > and
> > # has limited use.
> > #
> > # Default is 0 to disable saving the row cache.
> > row_cache_save_period: 0
> > 
> > # Number of keys from the row cache to save
> > # Disabled by default, meaning all keys are going to be saved
> > # row_cache_keys_to_save: 100
> > 
> > # Maximum size of the counter cache in memory.
> > #
> > # Counter cache helps to reduce counter locks' contention for hot counter 
> > cells.
> > # In case of RF = 1 a counter cache hit will cause Cassandra to skip the 
> > read before
> > # write entirely. With RF > 1 a counter cache hit will still help to reduce 
> > the duration
> > # of the lock hold, helping with hot counter cell updates, but will not 
> > allow skipping
> > # the read entirely. Only the local (clock, count) tuple of a counter cell 
> > is kept
> > # in memory, not the whole counter, so it's relatively cheap.
> > #
> > # NOTE: if you reduce the size, you may not get you hottest keys loaded on 
> > startup.
> > #
> > # Default value is empty to make it "auto" (min(2.5% of Heap (in MB), 
> > 50MB)). Set to 0 to disable counter cache.
> > # NOTE: if you perform counter deletes and rely on low gcgs, you should 
> > disable the counter cache.
> > counter_cache_size_in_mb:
> > 
> > # Duration in seconds after which Cassandra should
> > # save the counter cache (keys only). Caches are saved to 
> > saved_caches_directory as
> > # specified in this configuration file.
> > #
> > # Default is 7200 or 2 hours.
> > counter_cache_save_period: 7200
> > 
> > # Number of keys from the counter cache to save
> > # Disabled by default, meaning all keys are going to be saved
> > # counter_cache_keys_to_save: 100
> > 
> > # The off-heap memory allocator.  Affects storage engine metadata as
> > # well as caches.  Experiments show that JEMAlloc saves some memory
> > # than the native GCC allocator (i.e., JEMalloc is more
> > # fragmentation-resistant).
> > #
> > # Supported values are: NativeAllocator, JEMallocAllocator
> > #
> > # If you intend to use JEMallocAllocator you have to install JEMalloc as 
> > library and
> > # modify cassandra-env.sh as directed in the file.
> > #
> > # Defaults to NativeAllocator
> > # memory_allocator: NativeAllocator
> > 
> > # saved caches
> > # If not set, the default directory is $CASSANDRA_HOME/data/saved_caches.
> > saved_caches_directory: /data/cassandra/saved_caches
> > 
> > # commitlog_sync may be either "periodic" or "batch."
> > #
> > # When in batch mode, Cassandra won't ack writes until the commit log
> > # has been fsynced to disk.  It will wait
> > # commitlog_sync_batch_window_in_ms milliseconds between fsyncs.
> > # This window should be kept short because the writer threads will
> > # be unable to do extra work while waiting.  (You may need to increase
> > # concurrent_writes for the same reason.)
> > #
> > # commitlog_sync: batch
> > # commitlog_sync_batch_window_in_ms: 2
> > #
> > # the other option is "periodic" where writes may be acked immediately
> > # and the CommitLog is simply synced every commitlog_sync_period_in_ms
> > # milliseconds.
> > commitlog_sync: periodic
> > commitlog_sync_period_in_ms: 10000
> > 
> > # The size of the individual commitlog file segments.  A commitlog
> > # segment may be archived, deleted, or recycled once all the data
> > # in it (potentially from each columnfamily in the system) has been
> > # flushed to sstables.
> > #
> > # The default size is 32, which is almost always fine, but if you are
> > # archiving commitlog segments (see commitlog_archiving.properties),
> > # then you probably want a finer granularity of archiving; 8 or 16 MB
> > # is reasonable.
> > commitlog_segment_size_in_mb: 32
> > 
> > # Reuse commit log files when possible. The default is false, and this
> > # feature will be removed entirely in future versions of Cassandra.
> > #commitlog_segment_recycling: false
> > 
> > # any class that implements the SeedProvider interface and has a
> > # constructor that takes a Map<String, String> of parameters will do.
> > seed_provider:
> >    # Addresses of hosts that are deemed contact points.
> >    # Cassandra nodes use this list of hosts to find each other and learn
> >    # the topology of the ring.  You must change this if you are running
> >    # multiple nodes!
> >    - class_name: org.apache.cassandra.locator.SimpleSeedProvider
> >      parameters:
> >          # seeds is actually a comma-delimited list of addresses.
> >          # Ex: "<ip1>,<ip2>,<ip3>"
> >          - seeds: "xx, yy, zz"
> > 
> > # For workloads with more data than can fit in memory, Cassandra's
> > # bottleneck will be reads that need to fetch data from
> > # disk. "concurrent_reads" should be set to (16 * number_of_drives) in
> > # order to allow the operations to enqueue low enough in the stack
> > # that the OS and drives can reorder them. Same applies to
> > # "concurrent_counter_writes", since counter writes read the current
> > # values before incrementing and writing them back.
> > #
> > # On the other hand, since writes are almost never IO bound, the ideal
> > # number of "concurrent_writes" is dependent on the number of cores in
> > # your system; (8 * number_of_cores) is a good rule of thumb.
> > concurrent_reads: 32
> > concurrent_writes: 32
> > concurrent_counter_writes: 32
> > 
> > # Total memory to use for sstable-reading buffers.  Defaults to
> > # the smaller of 1/4 of heap or 512MB.
> > # file_cache_size_in_mb: 512
> > 
> > # Total permitted memory to use for memtables. Cassandra will stop
> > # accepting writes when the limit is exceeded until a flush completes,
> > # and will trigger a flush based on memtable_cleanup_threshold
> > # If omitted, Cassandra will set both to 1/4 the size of the heap.
> > # memtable_heap_space_in_mb: 2048
> > # memtable_offheap_space_in_mb: 2048
> > 
> > # Ratio of occupied non-flushing memtable size to total permitted size
> > # that will trigger a flush of the largest memtable.  Lager mct will
> > # mean larger flushes and hence less compaction, but also less concurrent
> > # flush activity which can make it difficult to keep your disks fed
> > # under heavy write load.
> > #
> > # memtable_cleanup_threshold defaults to 1 / (memtable_flush_writers + 1)
> > # memtable_cleanup_threshold: 0.11
> > 
> > # Specify the way Cassandra allocates and manages memtable memory.
> > # Options are:
> > #   heap_buffers:    on heap nio buffers
> > #   offheap_buffers: off heap (direct) nio buffers
> > #   offheap_objects: native memory, eliminating nio buffer heap overhead
> > #memtable_allocation_type: heap_buffers
> > memtable_allocation_type: offheap_objects
> > 
> > # Total space to use for commitlogs.  Since commitlog segments are
> > # mmapped, and hence use up address space, the default size is 32
> > # on 32-bit JVMs, and 8192 on 64-bit JVMs.
> > #
> > # If space gets above this value (it will round up to the next nearest
> > # segment multiple), Cassandra will flush every dirty CF in the oldest
> > # segment and remove it.  So a small total commitlog space will tend
> > # to cause more flush activity on less-active columnfamilies.
> > # commitlog_total_space_in_mb: 8192
> > 
> > # This sets the amount of memtable flush writer threads.  These will
> > # be blocked by disk io, and each one will hold a memtable in memory
> > # while blocked.
> > #
> > # memtable_flush_writers defaults to the smaller of (number of disks,
> > # number of cores), with a minimum of 2 and a maximum of 8.
> > #
> > # If your data directories are backed by SSD, you should increase this
> > # to the number of cores.
> > #memtable_flush_writers: 8
> > 
> > # A fixed memory pool size in MB for for SSTable index summaries. If left
> > # empty, this will default to 5% of the heap size. If the memory usage of
> > # all index summaries exceeds this limit, SSTables with low read rates will
> > # shrink their index summaries in order to meet this limit.  However, this
> > # is a best-effort process. In extreme conditions Cassandra may need to use
> > # more than this amount of memory.
> > index_summary_capacity_in_mb:
> > 
> > # How frequently index summaries should be resampled.  This is done
> > # periodically to redistribute memory from the fixed-size pool to sstables
> > # proportional their recent read rates.  Setting to -1 will disable this
> > # process, leaving existing index summaries at their current sampling level.
> > index_summary_resize_interval_in_minutes: 60
> > 
> > # Whether to, when doing sequential writing, fsync() at intervals in
> > # order to force the operating system to flush the dirty
> > # buffers. Enable this to avoid sudden dirty buffer flushing from
> > # impacting read latencies. Almost always a good idea on SSDs; not
> > # necessarily on platters.
> > trickle_fsync: true
> > trickle_fsync_interval_in_kb: 10240
> > 
> > # TCP port, for commands and data
> > # For security reasons, you should not expose this port to the internet.  
> > Firewall it if needed.
> > storage_port: 7000
> > 
> > # SSL port, for encrypted communication.  Unused unless enabled in
> > # encryption_options
> > # For security reasons, you should not expose this port to the internet.  
> > Firewall it if needed.
> > ssl_storage_port: 7001
> > 
> > # Address or interface to bind to and tell other Cassandra nodes to connect 
> > to.
> > # You _must_ change this if you want multiple nodes to be able to 
> > communicate!
> > #
> > # Set listen_address OR listen_interface, not both. Interfaces must 
> > correspond
> > # to a single address, IP aliasing is not supported.
> > #
> > # Leaving it blank leaves it up to InetAddress.getLocalHost(). This
> > # will always do the Right Thing _if_ the node is properly configured
> > # (hostname, name resolution, etc), and the Right Thing is to use the
> > # address associated with the hostname (it might not be).
> > #
> > # Setting listen_address to 0.0.0.0 is always wrong.
> > #
> > # If you choose to specify the interface by name and the interface has an 
> > ipv4 and an ipv6 address
> > # you can specify which should be chosen using 
> > listen_interface_prefer_ipv6. If false the first ipv4
> > # address will be used. If true the first ipv6 address will be used. 
> > Defaults to false preferring
> > # ipv4. If there is only one address it will be selected regardless of 
> > ipv4/ipv6.
> > listen_address: xxx
> > # listen_interface: eth0
> > # listen_interface_prefer_ipv6: false
> > 
> > # Address to broadcast to other Cassandra nodes
> > # Leaving this blank will set it to the same value as listen_address
> > # broadcast_address: 1.2.3.4
> > 
> > # Internode authentication backend, implementing IInternodeAuthenticator;
> > # used to allow/disallow connections from peer nodes.
> > # internode_authenticator: 
> > org.apache.cassandra.auth.AllowAllInternodeAuthenticator
> > 
> > # Whether to start the native transport server.
> > # Please note that the address on which the native transport is bound is the
> > # same as the rpc_address. The port however is different and specified 
> > below.
> > start_native_transport: true
> > # port for the CQL native transport to listen for clients on
> > # For security reasons, you should not expose this port to the internet.  
> > Firewall it if needed.
> > native_transport_port: 9042
> > # The maximum threads for handling requests when the native transport is 
> > used.
> > # This is similar to rpc_max_threads though the default differs slightly 
> > (and
> > # there is no native_transport_min_threads, idle threads will always be 
> > stopped
> > # after 30 seconds).
> > # native_transport_max_threads: 128
> > #
> > # The maximum size of allowed frame. Frame (requests) larger than this will
> > # be rejected as invalid. The default is 256MB.
> > # native_transport_max_frame_size_in_mb: 256
> > 
> > # The maximum number of concurrent client connections.
> > # The default is -1, which means unlimited.
> > # native_transport_max_concurrent_connections: -1
> > 
> > # The maximum number of concurrent client connections per source ip.
> > # The default is -1, which means unlimited.
> > # native_transport_max_concurrent_connections_per_ip: -1
> > 
> > # Whether to start the thrift rpc server.
> > start_rpc: true
> > 
> > # The address or interface to bind the Thrift RPC service and native 
> > transport
> > # server to.
> > #
> > # Set rpc_address OR rpc_interface, not both. Interfaces must correspond
> > # to a single address, IP aliasing is not supported.
> > #
> > # Leaving rpc_address blank has the same effect as on listen_address
> > # (i.e. it will be based on the configured hostname of the node).
> > #
> > # Note that unlike listen_address, you can specify 0.0.0.0, but you must 
> > also
> > # set broadcast_rpc_address to a value other than 0.0.0.0.
> > #
> > # For security reasons, you should not expose this port to the internet.  
> > Firewall it if needed.
> > #
> > # If you choose to specify the interface by name and the interface has an 
> > ipv4 and an ipv6 address
> > # you can specify which should be chosen using rpc_interface_prefer_ipv6. 
> > If false the first ipv4
> > # address will be used. If true the first ipv6 address will be used. 
> > Defaults to false preferring
> > # ipv4. If there is only one address it will be selected regardless of 
> > ipv4/ipv6.
> > rpc_address: xxx
> > # rpc_interface: eth1
> > # rpc_interface_prefer_ipv6: false
> > 
> > # port for Thrift to listen for clients on
> > rpc_port: 9160
> > 
> > # RPC address to broadcast to drivers and other Cassandra nodes. This cannot
> > # be set to 0.0.0.0. If left blank, this will be set to the value of
> > # rpc_address. If rpc_address is set to 0.0.0.0, broadcast_rpc_address must
> > # be set.
> > # broadcast_rpc_address: 1.2.3.4
> > 
> > # enable or disable keepalive on rpc/native connections
> > rpc_keepalive: true
> > 
> > # Cassandra provides two out-of-the-box options for the RPC Server:
> > #
> > # sync  -> One thread per thrift connection. For a very large number of 
> > clients, memory
> > #          will be your limiting factor. On a 64 bit JVM, 180KB is the 
> > minimum stack size
> > #          per thread, and that will correspond to your use of virtual 
> > memory (but physical memory
> > #          may be limited depending on use of stack space).
> > #
> > # hsha  -> Stands for "half synchronous, half asynchronous." All thrift 
> > clients are handled
> > #          asynchronously using a small number of threads that does not 
> > vary with the amount
> > #          of thrift clients (and thus scales well to many clients). The 
> > rpc requests are still
> > #          synchronous (one thread per active request). If hsha is selected 
> > then it is essential
> > #          that rpc_max_threads is changed from the default value of 
> > unlimited.
> > #
> > # The default is sync because on Windows hsha is about 30% slower.  On 
> > Linux,
> > # sync/hsha performance is about the same, with hsha of course using less 
> > memory.
> > #
> > # Alternatively,  can provide your own RPC server by providing the 
> > fully-qualified class name
> > # of an o.a.c.t.TServerFactory that can create an instance of it.
> > rpc_server_type: sync
> > 
> > # Uncomment rpc_min|max_thread to set request pool size limits.
> > #
> > # Regardless of your choice of RPC server (see above), the number of 
> > maximum requests in the
> > # RPC thread pool dictates how many concurrent requests are possible (but 
> > if you are using the sync
> > # RPC server, it also dictates the number of clients that can be connected 
> > at all).
> > #
> > # The default is unlimited and thus provides no protection against clients 
> > overwhelming the server. You are
> > # encouraged to set a maximum that makes sense for you in production, but 
> > do keep in mind that
> > # rpc_max_threads represents the maximum number of client requests this 
> > server may execute concurrently.
> > #
> > # rpc_min_threads: 16
> > # rpc_max_threads: 2048
> > 
> > # uncomment to set socket buffer sizes on rpc connections
> > # rpc_send_buff_size_in_bytes:
> > # rpc_recv_buff_size_in_bytes:
> > 
> > # Uncomment to set socket buffer size for internode communication
> > # Note that when setting this, the buffer size is limited by 
> > net.core.wmem_max
> > # and when not setting it it is defined by net.ipv4.tcp_wmem
> > # See:
> > # /proc/sys/net/core/wmem_max
> > # /proc/sys/net/core/rmem_max
> > # /proc/sys/net/ipv4/tcp_wmem
> > # /proc/sys/net/ipv4/tcp_wmem
> > # and: man tcp
> > # internode_send_buff_size_in_bytes:
> > # internode_recv_buff_size_in_bytes:
> > 
> > # Frame size for thrift (maximum message length).
> > thrift_framed_transport_size_in_mb: 15
> > 
> > # Set to true to have Cassandra create a hard link to each sstable
> > # flushed or streamed locally in a backups/ subdirectory of the
> > # keyspace data.  Removing these links is the operator's
> > # responsibility.
> > incremental_backups: false
> > 
> > # Whether or not to take a snapshot before each compaction.  Be
> > # careful using this option, since Cassandra won't clean up the
> > # snapshots for you.  Mostly useful if you're paranoid when there
> > # is a data format change.
> > snapshot_before_compaction: false
> > 
> > # Whether or not a snapshot is taken of the data before keyspace truncation
> > # or dropping of column families. The STRONGLY advised default of true
> > # should be used to provide data safety. If you set this flag to false, you 
> > will
> > # lose data on truncation or drop.
> > auto_snapshot: true
> > 
> > # When executing a scan, within or across a partition, we need to keep the
> > # tombstones seen in memory so we can return them to the coordinator, which
> > # will use them to make sure other replicas also know about the deleted 
> > rows.
> > # With workloads that generate a lot of tombstones, this can cause 
> > performance
> > # problems and even exaust the server heap.
> > # 
> > (http://www.datastax.com/dev/blog/cassandra-anti-patterns-queues-and-queue-like-datasets)
> > # Adjust the thresholds here if you understand the dangers and want to
> > # scan more tombstones anyway.  These thresholds may also be adjusted at 
> > runtime
> > # using the StorageService mbean.
> > tombstone_warn_threshold: 10000
> > tombstone_failure_threshold: 1000000
> > 
> > # Granularity of the collation index of rows within a partition.
> > # Increase if your rows are large, or if you have a very large
> > # number of rows per partition.  The competing goals are these:
> > #   1) a smaller granularity means more index entries are generated
> > #      and looking up rows withing the partition by collation column
> > #      is faster
> > #   2) but, Cassandra will keep the collation index in memory for hot
> > #      rows (as part of the key cache), so a larger granularity means
> > #      you can cache more hot rows
> > column_index_size_in_kb: 64
> > 
> > 
> > # Log WARN on any batch size exceeding this value. 5kb per batch by default.
> > # Caution should be taken on increasing the size of this threshold as it 
> > can lead to node instability.
> > batch_size_warn_threshold_in_kb: 50
> > 
> > # Number of simultaneous compactions to allow, NOT including
> > # validation "compactions" for anti-entropy repair.  Simultaneous
> > # compactions can help preserve read performance in a mixed read/write
> > # workload, by mitigating the tendency of small sstables to accumulate
> > # during a single long running compactions. The default is usually
> > # fine and if you experience problems with compaction running too
> > # slowly or too fast, you should look at
> > # compaction_throughput_mb_per_sec first.
> > #
> > # concurrent_compactors defaults to the smaller of (number of disks,
> > # number of cores), with a minimum of 2 and a maximum of 8.
> > #
> > # If your data directories are backed by SSD, you should increase this
> > # to the number of cores.
> > concurrent_compactors: 4
> > 
> > # Throttles compaction to the given total throughput across the entire
> > # system. The faster you insert data, the faster you need to compact in
> > # order to keep the sstable count down, but in general, setting this to
> > # 16 to 32 times the rate you are inserting data is more than sufficient.
> > # Setting this to 0 disables throttling. Note that this account for all 
> > types
> > # of compaction, including validation compaction.
> > #compaction_throughput_mb_per_sec: 16
> > compaction_throughput_mb_per_sec: 64
> > 
> > # Log a warning when compacting partitions larger than this value
> > compaction_large_partition_warning_threshold_mb: 100
> > 
> > # When compacting, the replacement sstable(s) can be opened before they
> > # are completely written, and used in place of the prior sstables for
> > # any range that has been written. This helps to smoothly transfer reads
> > # between the sstables, reducing page cache churn and keeping hot rows hot
> > sstable_preemptive_open_interval_in_mb: 50
> > 
> > # Throttles all outbound streaming file transfers on this node to the
> > # given total throughput in Mbps. This is necessary because Cassandra does
> > # mostly sequential IO when streaming data during bootstrap or repair, which
> > # can lead to saturating the network connection and degrading rpc 
> > performance.
> > # When unset, the default is 200 Mbps or 25 MB/s.
> > # stream_throughput_outbound_megabits_per_sec: 200
> > 
> > # Throttles all streaming file transfer between the datacenters,
> > # this setting allows users to throttle inter dc stream throughput in 
> > addition
> > # to throttling all network stream traffic as configured with
> > # stream_throughput_outbound_megabits_per_sec
> > # When unset, the default is 200 Mbps or 25 MB/s
> > inter_dc_stream_throughput_outbound_megabits_per_sec: 400
> > 
> > # How long the coordinator should wait for read operations to complete
> > read_request_timeout_in_ms: 10000
> > # How long the coordinator should wait for seq or index scans to complete
> > range_request_timeout_in_ms: 10000
> > # How long the coordinator should wait for writes to complete
> > write_request_timeout_in_ms: 10000
> > # How long the coordinator should wait for counter writes to complete
> > counter_write_request_timeout_in_ms: 5000
> > # How long a coordinator should continue to retry a CAS operation
> > # that contends with other proposals for the same row
> > cas_contention_timeout_in_ms: 1000
> > # How long the coordinator should wait for truncates to complete
> > # (This can be much longer, because unless auto_snapshot is disabled
> > # we need to flush first so we can snapshot before removing the data.)
> > truncate_request_timeout_in_ms: 60000
> > # The default timeout for other, miscellaneous operations
> > request_timeout_in_ms: 10000
> > 
> > # Enable operation timeout information exchange between nodes to accurately
> > # measure request timeouts.  If disabled, replicas will assume that requests
> > # were forwarded to them instantly by the coordinator, which means that
> > # under overload conditions we will waste that much extra time processing
> > # already-timed-out requests.
> > #
> > # Warning: before enabling this property make sure to ntp is installed
> > # and the times are synchronized between the nodes.
> > cross_node_timeout: false
> > 
> > # Enable socket timeout for streaming operation.
> > # When a timeout occurs during streaming, streaming is retried from the 
> > start
> > # of the current file. This _can_ involve re-streaming an important amount 
> > of
> > # data, so you should avoid setting the value too low.
> > # Default value is 3600000, which means streams timeout after an hour.
> > # streaming_socket_timeout_in_ms: 3600000
> > 
> > # phi value that must be reached for a host to be marked down.
> > # most users should never need to adjust this.
> > # phi_convict_threshold: 8
> > phi_convict_threshold: 16
> > 
> > # endpoint_snitch -- Set this to a class that implements
> > # IEndpointSnitch.  The snitch has two functions:
> > # - it teaches Cassandra enough about your network topology to route
> > #   requests efficiently
> > # - it allows Cassandra to spread replicas around your cluster to avoid
> > #   correlated failures. It does this by grouping machines into
> > #   "datacenters" and "racks."  Cassandra will do its best not to have
> > #   more than one replica on the same "rack" (which may not actually
> > #   be a physical location)
> > #
> > # IF YOU CHANGE THE SNITCH AFTER DATA IS INSERTED INTO THE CLUSTER,
> > # YOU MUST RUN A FULL REPAIR, SINCE THE SNITCH AFFECTS WHERE REPLICAS
> > # ARE PLACED.
> > #
> > # IF THE RACK A REPLICA IS PLACED IN CHANGES AFTER THE REPLICA HAS BEEN
> > # ADDED TO A RING, THE NODE MUST BE DECOMMISSIONED AND REBOOTSTRAPPED.
> > #
> > # Out of the box, Cassandra provides
> > #  - SimpleSnitch:
> > #    Treats Strategy order as proximity. This can improve cache
> > #    locality when disabling read repair.  Only appropriate for
> > #    single-datacenter deployments.
> > #  - GossipingPropertyFileSnitch
> > #    This should be your go-to snitch for production use.  The rack
> > #    and datacenter for the local node are defined in
> > #    cassandra-rackdc.properties and propagated to other nodes via
> > #    gossip.  If cassandra-topology.properties exists, it is used as a
> > #    fallback, allowing migration from the PropertyFileSnitch.
> > #  - PropertyFileSnitch:
> > #    Proximity is determined by rack and data center, which are
> > #    explicitly configured in cassandra-topology.properties.
> > #  - Ec2Snitch:
> > #    Appropriate for EC2 deployments in a single Region. Loads Region
> > #    and Availability Zone information from the EC2 API. The Region is
> > #    treated as the datacenter, and the Availability Zone as the rack.
> > #    Only private IPs are used, so this will not work across multiple
> > #    Regions.
> > #  - Ec2MultiRegionSnitch:
> > #    Uses public IPs as broadcast_address to allow cross-region
> > #    connectivity.  (Thus, you should set seed addresses to the public
> > #    IP as well.) You will need to open the storage_port or
> > #    ssl_storage_port on the public IP firewall.  (For intra-Region
> > #    traffic, Cassandra will switch to the private IP after
> > #    establishing a connection.)
> > #  - RackInferringSnitch:
> > #    Proximity is determined by rack and data center, which are
> > #    assumed to correspond to the 3rd and 2nd octet of each node's IP
> > #    address, respectively.  Unless this happens to match your
> > #    deployment conventions, this is best used as an example of
> > #    writing a custom Snitch class and is provided in that spirit.
> > #
> > # You can use a custom Snitch by setting this to the full class name
> > # of the snitch, which will be assumed to be on your classpath.
> > #endpoint_snitch: SimpleSnitch
> > endpoint_snitch: com.ddd.cassandra.OldEC2Snitch
> > 
> > # controls how often to perform the more expensive part of host score
> > # calculation
> > dynamic_snitch_update_interval_in_ms: 100
> > # controls how often to reset all host scores, allowing a bad host to
> > # possibly recover
> > dynamic_snitch_reset_interval_in_ms: 600000
> > # if set greater than zero and read_repair_chance is < 1.0, this will allow
> > # 'pinning' of replicas to hosts in order to increase cache capacity.
> > # The badness threshold will control how much worse the pinned host has to 
> > be
> > # before the dynamic snitch will prefer other replicas over it.  This is
> > # expressed as a double which represents a percentage.  Thus, a value of
> > # 0.2 means Cassandra would continue to prefer the static snitch values
> > # until the pinned host was 20% worse than the fastest.
> > dynamic_snitch_badness_threshold: 0.1
> > 
> > # request_scheduler -- Set this to a class that implements
> > # RequestScheduler, which will schedule incoming client requests
> > # according to the specific policy. This is useful for multi-tenancy
> > # with a single Cassandra cluster.
> > # NOTE: This is specifically for requests from the client and does
> > # not affect inter node communication.
> > # org.apache.cassandra.scheduler.NoScheduler - No scheduling takes place
> > # org.apache.cassandra.scheduler.RoundRobinScheduler - Round robin of
> > # client requests to a node with a separate queue for each
> > # request_scheduler_id. The scheduler is further customized by
> > # request_scheduler_options as described below.
> > request_scheduler: org.apache.cassandra.scheduler.NoScheduler
> > 
> > # Scheduler Options vary based on the type of scheduler
> > # NoScheduler - Has no options
> > # RoundRobin
> > #  - throttle_limit -- The throttle_limit is the number of in-flight
> > #                      requests per client.  Requests beyond
> > #                      that limit are queued up until
> > #                      running requests can complete.
> > #                      The value of 80 here is twice the number of
> > #                      concurrent_reads + concurrent_writes.
> > #  - default_weight -- default_weight is optional and allows for
> > #                      overriding the default which is 1.
> > #  - weights -- Weights are optional and will default to 1 or the
> > #               overridden default_weight. The weight translates into how
> > #               many requests are handled during each turn of the
> > #               RoundRobin, based on the scheduler id.
> > #
> > # request_scheduler_options:
> > #    throttle_limit: 80
> > #    default_weight: 5
> > #    weights:
> > #      Keyspace1: 1
> > #      Keyspace2: 5
> > 
> > # request_scheduler_id -- An identifier based on which to perform
> > # the request scheduling. Currently the only valid option is keyspace.
> > # request_scheduler_id: keyspace
> > 
> > # internode_compression controls whether traffic between nodes is
> > # compressed.
> > # can be:  all  - all traffic is compressed
> > #          dc   - traffic between different datacenters is compressed
> > #          none - nothing is compressed.
> > internode_compression: all
> > 
> > # Enable or disable tcp_nodelay for inter-dc communication.
> > # Disabling it will result in larger (but fewer) network packets being sent,
> > # reducing overhead from the TCP protocol itself, at the cost of increasing
> > # latency if you block for cross-datacenter responses.
> > inter_dc_tcp_nodelay: false
> > 
> > # GC Pauses greater than gc_warn_threshold_in_ms will be logged at WARN 
> > level
> > # Adjust the threshold based on your application throughput requirement
> > # By default, Cassandra logs GC Pauses greater than 200 ms at INFO level
> > # gc_warn_threshold_in_ms: 1000
> > 
> > =================================
> > 
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: user-unsubscr...@cassandra.apache.org
> > For additional commands, e-mail: user-h...@cassandra.apache.org
> > 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: user-unsubscr...@cassandra.apache.org
> For additional commands, e-mail: user-h...@cassandra.apache.org
> 
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscr...@cassandra.apache.org
For additional commands, e-mail: user-h...@cassandra.apache.org