ecc.yml

#
# Copyright 2020 Telefonaktiebolaget LM Ericsson
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

### ecChronos configuration

## Connection
## Properties for connection to the local node
##
connection:
  cql:
    ##
    ## Host and port properties for CQL.
    ## Primarily used by the default connection provider
    ##
    host: localhost
    port: 9042
    ##
    ## Connection Timeout for a CQL attempt.
    ## Specify a time to wait for cassandra to come up.
    ## Connection is tried based on retry policy delay calculations. Each connection attempt will use the timeout to calculate CQL connection process delay.
    ##
    timeout:
      time: 60
      unit: seconds
    retryPolicy:
      ## Max number of attempts ecChronos will try to connect with Cassandra.
      maxAttempts: 5
      ## Delay use to wait between an attempt and another, this value will be multiplied by the current attempt count powered by two.
      ## If the current attempt is 4 and the default delay is 5 seconds, so ((4(attempt) x 2) x 5(default delay)) = 40 seconds.
      ## If the calculated delay is greater than maxDelay, maxDelay will be used instead of the calculated delay.
      delay: 5
      ## Maximum delay before the next connection attempt is made.
      ## Setting it as 0 will disable maxDelay and the delay interval will
      ## be calculated based on the attempt count and the default delay.
      maxDelay: 30
      unit: seconds
    ##
    ## The class used to provide CQL connections to Apache Cassandra.
    ## The default provider will be used unless another is specified.
    ##
    provider: com.ericsson.bss.cassandra.ecchronos.application.DefaultNativeConnectionProvider
    ##
    ## The class used to provide an SSL context to the NativeConnectionProvider.
    ## Extending this allows to manipulate the SSLEngine and SSLParameters.
    ##
    certificateHandler: com.ericsson.bss.cassandra.ecchronos.application.ReloadingCertificateHandler
    ##
    ## The class used to decorate CQL statements.
    ## The default no-op decorator will be used unless another is specified.
    ##
    decoratorClass: com.ericsson.bss.cassandra.ecchronos.application.NoopStatementDecorator
    ##
    ## Allow routing requests directly to a remote datacenter.
    ## This allows locks for other datacenters to be taken in that datacenter instead of via the local datacenter.
    ## If clients are prevented from connecting directly to Cassandra nodes in other sites this is not possible.
    ## If remote routing is disabled, instead SERIAL consistency will be used for those request.
    ##
    remoteRouting: true
  jmx:
    ##
    ## Host and port properties for JMX.
    ## Primarily used by the default connection provider.
    ##
    host: localhost
    port: 7199
    ##
    ## The class used to provide JMX connections to Apache Cassandra.
    ## The default provider will be used unless another is specified.
    ##
    provider: com.ericsson.bss.cassandra.ecchronos.application.DefaultJmxConnectionProvider

## Repair configuration
## This section defines default repair behavior for all tables.
##
repair:
  ##
  ## A class for providing repair configuration for tables.
  ## The default FileBasedRepairConfiguration uses a schedule.yml file to define per-table configurations.
  ##
  provider: com.ericsson.bss.cassandra.ecchronos.application.FileBasedRepairConfiguration
  ##
  ## How often repairs should be triggered for tables.
  ##
  interval:
    time: 7
    unit: days
  ##
  ## Initial delay for new tables. New tables are always assumed to have been repaired in the past by the interval.
  ## However, a delay can be set for the first repair. This will not affect subsequent repairs and defaults to one day.
  ##
  initial_delay:
    time: 1
    unit: days
  ##
  ## The unit of time granularity for priority calculation, can be HOURS, MINUTES, or SECONDS.
  ## This unit is used in the calculation of priority.
  ## Default is HOURS for backward compatibility.
  ## Ensure to pause repair operations prior to changing the granularity.
  ## Not doing so may lead to inconsistencies as some ecchronos instances
  ## could have different priorities compared to others for the same repair.
  ## Possible values are HOURS, MINUTES, or SECONDS.
  ##
  priority:
    granularity_unit: HOURS
  ##
  ## Specifies the type of lock to use for repairs.
  ## "vnode" will lock each node involved in a repair individually and increase the number of
  ## parallel repairs that can run in a single data center.
  ## "datacenter" will lock each data center involved in a repair and only allow a single repair per data center.
  ## "datacenter_and_vnode" will combine both options and allow a smooth transition between them without allowing
  ## multiple repairs to run concurrently on a single node.
  ##
  lock_type: vnode
  ##
  ## Alarms are triggered when tables have not been repaired for a long amount of time.
  ## The warning alarm is meant to indicate early that repairs are falling behind.
  ## The error alarm is meant to indicate that gc_grace has passed between repairs.
  ##
  ## With the defaults where repairs triggers once every 7 days for each table a warning alarm would be raised
  ## if the table has not been properly repaired within one full day.
  ##
  alarm:
    ##
    ## The class used for fault reporting
    ## The default LoggingFaultReporter will log when alarm is raised/ceased
    ##
    faultReporter: com.ericsson.bss.cassandra.ecchronos.fm.impl.LoggingFaultReporter
    ##
    ## If a table has not been repaired for the following duration an warning alarm will be raised.
    ## The schedule will be marked as late if the table has not been repaired within this interval.
    ##
    warn:
      time: 8
      unit: days
    ##
    ## If a table has not been repaired for the following duration an error alarm will be raised.
    ## The schedule will be marked as overdue if the table has not been repaired within this interval.
    ##
    error:
      time: 10
      unit: days
  ##
  ## Specifies the unwind ratio to smooth out the load that repairs generate.
  ## This value is a ratio between 0 -> 100% of the execution time of a repair session.
  ##
  ## 100% means that the executor will wait to run the next session for as long time as the previous session took.
  ## The 'unwind_ratio' setting configures the wait time between repair tasks as a proportion of the previous task's execution time.
  ##
  ## Examples:
  ## - unwind_ratio: 0
  ##   Explanation: No wait time between tasks. The next task starts immediately after the previous one finishes.
  ##   Total Repair Time: T1 (10s) + T2 (20s) = 30 seconds.
  ##
  ## - unwind_ratio: 1.0 (100%)
  ##   Explanation: The wait time after each task equals its duration.
  ##   Total Repair Time: T1 (10s + 10s wait) + T2 (20s + 20s wait) = 60 seconds.
  ##
  ## - unwind_ratio: 0.5 (50%)
  ##   Explanation: The wait time is half of the task's duration.
  ##   Total Repair Time: T1 (10s + 5s wait) + T2 (20s + 10s wait) = 45 seconds.
  ##
  ##  A higher 'unwind_ratio' reduces system load by adding longer waits, but increases total repair time.
  ##  A lower 'unwind_ratio' speeds up repairs but may increase system load.
  ##
  unwind_ratio: 0.0
  ##
  ## Specifies the lookback time for when the repair_history table is queried to get initial repair state at startup.
  ## The time should match the "expected TTL" of the system_distributed.repair_history table.
  ##
  history_lookback:
    time: 30
    unit: days
  ##
  ## Specifies a target for how much data each repair session should process.
  ## This is only supported if using 'vnode' as repair_type.
  ## This is an estimation assuming uniform data distribution among partition keys.
  ## The value should be either a number or a number with a unit of measurement:
  ## 12  (12 B)
  ## 12k (12 KiB)
  ## 12m (12 MiB)
  ## 12g (12 GiB)
  ##
  size_target:
  ##
  ## Specifies the repair history provider used to determine repair state.
  ## The "cassandra" provider uses the repair history generated by the database.
  ## The "upgrade" provider is an intermediate state reading history from "cassandra" and producing history for "ecc"
  ## The "ecc" provider maintains and uses an internal repair history in a dedicated table.
  ## The main context for the "ecc" provider is an environment where the ip address of nodes might change.
  ## Possible values are "ecc", "upgrade" and "cassandra".
  ##
  ## The keyspace parameter is only used by "ecc" and "upgrade" and points to the keyspace where the custom
  ## 'repair_history' table is located.
  ##
  history:
    provider: ecc
    keyspace: ecchronos
  ##
  ## Specifies if tables with TWCS (TimeWindowCompactionStrategy) should be ignored for repair
  ##
  ignore_twcs_tables: false
  ##
  ## Specifies the backoff time for a job.
  ## This is the time that the job will wait before trying to run again after failing.
  ##
  backoff:
    time: 30
    unit: MINUTES
  ##
  ## Specifies the default repair_type.
  ## Possible values are: vnode, parallel_vnode, incremental
  ## vnode = repair 1 vnode at a time (supports size_target to split the vnode further, in this case there will be 1 repair session per subrange)
  ## parallel_vnode = repair vnodes in parallel, this will combine vnodes into a single repair session per repair group
  ## incremental = repair vnodes incrementally (incremental repair)
  ##
  repair_type: vnode

statistics:
  enabled: true
  ##
  ## Decides how statistics should be exposed.
  ## If all reporting is disabled, the statistics will be disabled as well.
  ##
  reporting:
    jmx:
      enabled: true
      ##
      ## The metrics can be excluded on name and on tag values using quoted regular expressions.
      ## Exclusion on name should be done without the prefix.
      ## If an exclusion is without tags, then metric matching the name will be excluded.
      ## If both name and tags are specified, then the metric must match both to be excluded.
      ## If multiple tags are specified, the metric must match all tags to be excluded.
      ## By default, no metrics are excluded.
      ## For list of available metrics and tags refer to the documentation.
      ##
      excludedMetrics: []
    file:
      enabled: true
      ##
      ## The metrics can be excluded on name and on tag values using quoted regular expressions.
      ## Exclusion on name should be done without the prefix.
      ## If an exclusion is without tags, then metric matching the name will be excluded.
      ## If both name and tags are specified, then the metric must match both to be excluded.
      ## If multiple tags are specified, the metric must match all tags to be excluded.
      ## By default, no metrics are excluded.
      ## For list of available metrics and tags refer to the documentation.
      ##
      excludedMetrics: []
    http:
      enabled: true
      ##
      ## The metrics can be excluded on name and on tag values using quoted regular expressions.
      ## Exclusion on name should be done without the prefix.
      ## If an exclusion is without tags, then metric matching the name will be excluded.
      ## If both name and tags are specified, then the metric must match both to be excluded.
      ## If multiple tags are specified, the metric must match all tags to be excluded.
      ## By default, no metrics are excluded.
      ## For list of available metrics and tags refer to the documentation.
      ##
      excludedMetrics: []
  directory: ./statistics
  ##
  ## Prefix all metrics with below string
  ## The prefix cannot start or end with a dot or any other path separator.
  ##
  prefix: ''
  ##
  ## Number of repair failures before status logger logs metrics in debug logs
  ## The number is used to trigger a status once number of failures is breached in a time window mentioned below
  ##
  repair_failures_count: 5
  ##
  ## Time window over which to track repair failures in node for trigger status logger messages in debug log
  ##
  repair_failures_time_window:
      time: 30
      unit: minutes
  ##
  ## Trigger interval for metric inspection.
  ## This time should always be lesser than repair_failures_time_window
  ##
  trigger_interval_for_metric_inspection:
       time: 5
       unit: seconds

lock_factory:
  cas:
    ##
    ## The keyspace used for the CAS lock factory tables.
    ##
    keyspace: ecchronos
    ##
    ## The number of seconds until the lock failure cache expires.
    ## If an attempt to secure a lock is unsuccessful,
    ## all subsequent attempts will be failed until
    ## the cache expiration time is reached.
    ##
    cache_expiry_time_in_seconds: 30
    ##
    ## Allow to override consistency level for LWT (lightweight transactions). Possible values are:
    ## "DEFAULT" - Use consistency level based on remoteRouting.
    ## "SERIAL" - Use SERIAL consistency for LWT regardless of remoteRouting.
    ## "LOCAL" - Use LOCAL_SERIAL consistency for LWT regardless of remoteRouting.
    ##
    ## if you use remoteRouting: false and LOCAL then all locks will be taken locally
    ## in DC. I.e There's a risk that multiple nodes in different datacenters will be able to lock the
    ## same nodes causing multiple repairs on the same range/node at the same time.
    ##
    consistencySerial: "DEFAULT"

run_policy:
  time_based:
    ##
    ## The keyspace used for the time based run policy tables.
    ##
    keyspace: ecchronos

scheduler:
  ##
  ## Specifies the frequency the scheduler checks for work to be done
  ##
  frequency:
    time: 30
    unit: SECONDS

rest_server:
  ##
  ## The host and port used for the HTTP server
  ##
  host: localhost
  port: 8080