2.0.0

2.0 (2019/12/06 06:25 +00:00)

• 848e48b Add "force" flag to cluster delete REST endpoint, allowing registered clusters with schedules and repairs runs to be forcibly deleted. (@michaelsembwever)
• e13b34d Remove leader election in Reaper for schema migrations and instead rely on the cassandra-migration lib built in consensus. (@adejanovski)
• da5a058 Display available repair metrics in the UI (@adejanovski)
• 2243c73 Make docker image run as non-root (@mimarpe)
• edfd5a9 Added option REAPER_AUTH_ENABLED. (@andrejvanderzee)
• 9bb2878 Add switchable UI themes (@adejanovski)
  aa833d8) support use of Postgresql as storage backend for multiple reapers (@Saleil-Bhat)
• 0591b38 Make stream tracking work with 4.0 (@rzvoncek)
• b293160 Add Cassandra storage support for Diagnostic Events (@michaelsembwever)
• e65a47e Add JMX and REST support for diagnostic events (@spodkowinski)
• a04e7cf Fix NPE upon upgrade when last contact is null on existing clusters (@adejanovski)
• 22c7cfe Self register clusters in sidecar mode since the nodes are collocated with Reaper. (@adejanovski)
• a5fdac2 Simplify how RepairManager aborts/postpones segments, fixing how leader election happens and when and how it is correctly released. (@michaelsembwever)
• f01e914 Implement the sidecar mode (@adejanovski)
• f798244 Handle new messaging service verbs for dropped reads in Cassandra 4.0 (@michaelsembwever)
• fed72fe Request remote nodes metrics only if there are several reapers running (@adejanovski)
• 644553a Improves the preventation of multiple scheduled repair runs starting at the same time. (@michaelsembwever)
• 65909e3 The purge and abort segment REST endpoints are actions, not idempotent reads, so should be POST instead of GET method requests. (@michaelsembwever)
• e686c2f Don't start a segment repair if it is already started, running, or done. (@michaelsembwever)
• d7d045f Add automated retries when connecting to the storage backend (@adejanovski)
• f6c451b Encapsulate jmx interactions in a proxy class (@adejanovski)

1.4.7

1.4.7 (2019/10/16 13:18 +00:00)

• bb39388 Fix metric naming to allow long term tracking of repairs. (@adejanovski)
• 68f92a4 Cluster view UI reorg (@adejanovski)

1.4.5

1.4.5 (2019/08/23 15:36 +00:00)

• 8bcf6d3 Backport multiple DC UI fix to 1.4 (@michaelsembwever)
• 3dcb5d5 Fix the find command to correctly identify the reaper jar on all systems while excluding source/javadoc jars (@adejanovski)
• 1cb9075 Add authorization to rest endpoints based on their verbs (http method), add "operator" (read+write) and "user" (read-only) roles, let the existing admin user be an operator and add a new user "user" with the "user" role. (@michaelsembwever)
• 134c7a0 Don't include *-sources.jar into the classpath, and be a bit smarter in finding the jar file in a current project directory (regardless of the current folder in the project directory) (@michaelsembwever)
• 534f5aa Fix initialisation ordering of AppContext (@treuherz)
• a934aea Improve performance of JsonParseUtils (@michaelsembwever)

1.4.4

Change Log

1.4.4 (2019/05/28 09:21 +00:00)

• 47322c1 Fix for NPE when a segment is aborted (@adejanovski)

1.4.3

Change Log

1.4.3 (2019/05/21 00:43 +00:00)

• 91f6f51 Automatically run every Cassandra Migration class when upgrading (@michaelsembwever)
• 2b82937 Change UI run_repair API call to be more performant (@ossarga)
• 544de58 In SegmentRunner immediately release the segment lead once the segment repair is done. The intensity pause can happen afterwards. (@michaelsembwever)
• c6e1098 Fix (set to null) endTime in any repairRuns that are in PAUSED state (@michaelsembwever)
• 5cdd573 Fix RepairRun timestamps, honouring different run states. (@michaelsembwever)
• c0b4600 Auto scheduler encounters null pointer exception #298 (@michaelsembwever)
• cf34fbb Fix LongCollectionSqlType.getValue() (@mmonemali)
• b075113 ColumnFamilyStoreMBean.getCompactionParameters() is only available since Cassandra 2.1 (@rborer)

1.4.1

Changelog

• 7f46e20 Add newline at the end of shiro.ini to fix REAPER_AUTH_USER (@viossat)
• b7d6593 In NodesStatus when parsing gossipinfo status, also exclude those nodes that have been "removed", as well as those that are "left". (@michaelsembwever)
• 0a2c11f Fix for #613 when driver class is missing from config (@adejanovski)