cassandra-reaper

1.1.0 (2018/02/28 19:25 +00:00)

• e5519a7 updating metrics for repairProgress,segmentsDone,segmentsTotal and added keyspace to metric (@kingm5)
• 3cf121d Prevent stuck segment runner issues on uncatched exceptions (@adejanovski)
• f7a148d Remove unnecessary postpones when a segment isn't allowed to start (@adejanovski)
• bd17578 Ninja fix for error trace overload (@adejanovski)
• 6f3f3fa Initialize JMX connections on startup only when necessary. (@adejanovski)
• 00c2cc7 Fix issue with poorly registered metrics upon node restarts (@adejanovski)
• 07a897c Ninja fix for segment abort (due to a bad rebase) (@adejanovski)
• b9d8681 Switch to re-usable JMX connections instead of re-creating one each time it's needed. (@adejanovski)
• ca2341a UI update for ordering fixes (@adejanovski)
• ea1a0a0 Fix orderings of cluster names, repair runs and schedules (@adejanovski)
• 283f7e2 Add Maven uploads to Bintray (@adejanovski)
• 39e8cab turn on gpg signing of uploaded packages (@adejanovski)
• 6e2afd3 Ninja fix for hardcoded urls (@adejanovski)
• a07d2e6 UI update for segment display/abort (@adejanovski)
• 263f4ec Add ability to list and abort segments (@adejanovski)
• d30d79f fix build by using fpm directly instead of docker-compose (@adejanovski)
• 35694c3 Handle cases where a segment reaches timeout without being in either RUNNING or DONE state. (@adejanovski)
• 0091a4b Push all packages to Bintray on the master branch (beta) and tags (releases). (@adejanovski)
• 65fd58f Prevent crashing the scheduler for RepairManager.resumePendingRepairRuns() in case of unexpected exception. (@adejanovski)
• 07eadc3 added 1st set of metrics suggested (@debswade)
• 67d6721 Another fix for IllegalStateException: startTime must be set if segment is RUNNING or DONE (@michaelsembwever)
• ff0214f Ninja fix for issue #288 (@adejanovski)
• 13524bc Fix computing too many segments on high clusters with high number of vnodes. (@adejanovski)
• 5a4bfe9 NodeStatus: Handle hostnames in endpointsStatus output (@tedoc2000)
• 6f71b67 Implemented PUT method for change intensity (@kipwoker)
• 5b563db Supported change intensity in web app (@kipwoker)
• 6d06781 Only disable RepairSegment precondition checks when JVM is running with system property "-Dreaper.disableSegmentChecks=true" (@michaelsembwever)

• 5c9ca98 Ugly shameful fix for the endTime/startTime infamous bug (@adejanovski)

• e3b2cd7 Fix for Elassandra's gossip info output parsing (@adejanovski)
• 5a14d92 Duplicate the Migration009 script so it can run again to fix missing RepairSegment startTimes (@michaelsembwever)
• b56ce12 breakup the SegmentRunner.runRepair(..) method, the bulk of the body going into SegmentRunner.processTriggeredSegment(..) (@michaelsembwever)
• f9ea67e if endTime is set, so must startTime be set #281 (@michaelsembwever)

1.0.0 (2017/11/14 20:58 +00:00)

• ba9ca5f Release 1.0.0 final (@adejanovski)

1.0.0-RC3 (2017/11/13 10:01 +00:00)

• 91b2737 Ninja fix to have Reaper work with 2.0 again (@adejanovski)

1.0.0-RC2 (2017/11/09 16:50 +00:00)

• 02e53d0 Should resolve current Docker image issues (@joaquincasares)
• d75ba7b Don't log every jmx connection failure within JmxProxyImpl. The thrown exception is logged appropriately when caught. (@michaelsembwever)
• 5d3d469 enable assertions by default when running Reaper. (@michaelsembwever)
• 257bab2 Fix typo: endabled -> enabled (@SaaldjorMike)

1.0.0-RC1 (2017/11/03 11:44 +00:00)

• dc9c00d Pulled back the "database" setting for storage type to ensure backwards compatibility (@adejanovski)

• 92a120c Split 'database' configuration setting (@ossarga)

• 1c64c1a Minimal fix for test failure as in described in https://github.com/thelastpickle/cassandra-reaper/issues/266 (@michaelsembwever)

• 42c4c14 Add a doc page about multi DC clusters (@adejanovski)

• 13d9246 Merge branch 'master' of https://github.com/thelastpickle/cassandra-reaper (@adejanovski)

• 10d3b02 Fix maven ui build with absolute path in pom instead of relative (@adejanovski)

• 8241eab fixed ignore for packages (@rustyrazorblade)

• c0a4b7a Add version number in the UI (@adejanovski)

• c85e43c Automating release process using rultor (@velo)

• 619b28a Correct cassandra storage table properties. Always enable the key cache, and disable all read_repairs. (@michaelsembwever)

• e5dbedf Remove localMode configuration and logic, as it's not actually implemented yet. (@michaelsembwever)

• 91e5464 Fix repair_segment table start_time/end_time/state consistency (@adejanovski)

• d6e5a10 Remove NULLs in RepairSegments stored in Cassandra. (@michaelsembwever)

• 20f5558 Fix issue with upgrades on postgres and h2 when coming from pre-0.8 versions (@adejanovski)

• #256 Add metrics for repair progress + time since last repair. Fixes #207. (@thelastpickle)

• 2decdde dawdle duckling to 41ecc2b : node_metrics_v1.minute is a bigint (long) (@michaelsembwever)

• ee37c5a Add metrics for repair progress + time since last repair. Fixes #207. (@rzvoncek)

• 41ecc2b Add request mechanism for metrics to allow datacenterAvailability = EACH to work properly on large clusters (@adejanovski)

• ad41d64 Refactor test packages to match the runtime packages. Reduce class and methods visibilities where possible. (@michaelsembwever)

• d6a8eb4 Cleanup naming: use "node" consistently, in place of "host". (@michaelsembwever)

• ff19c38 Make RepairManager and Heart hold the AppContext as a field, rather than a parameter to all methods. (@michaelsembwever)

• 83d5528 Break apart CommonTools. Its methods put it correct peer classes, or into new service classes. (@michaelsembwever)

• c572e57 Give datacenterAvailability = EACH better chances to work, through requesting for and updating node metrics for running repairs, as part of heartbeat. (@michaelsembwever)

• 2b806d3 Unit tests for Heart (@michaelsembwever)

• 27b8038 set heap sizes for travis builds to the same as circleci (which displays better stability) (@michaelsembwever)

• f188f92 Rebuilt docs (@ossarga)

• 6f10a47 Updated configuration docs and Docker envs (@ossarga)

• 40e852d Fix minor UI issues with autocomplete fields (@adejanovski)

• 8bf9aae UI changes for the new segment count per node (@adejanovski)

• f3de7a3 Switch from a global segment count to a segment count per node (@adejanovski)

• a8aae66 Fix browser cache issues when the UI evolves. (@adejanovski)

• e717ad8 For Cassandra storage minimise the number of tombstones (NULL entries) in the repair_run table. (@michaelsembwever)

• 22afbdb Fix JmxConnectionFactory.connectAny(..) to loop a second time trying all hosts regardless of past successes/failures. (@michaelsembwever)

• e899534 Doc fix for correct docker build commands (@michaelsembwever)

• #228 Create CODE_OF_CONDUCT.md (#228) (@michaelsembwever)

• 9227f61 Fix NPE in CassandraStorage, when a read times out and the statement is null. (@michaelsembwever)

• 7b56121 Fix assertions around schedules, as the jersey response need not be successful, and then a correct response has to be waited for. (@michaelsembwever)

• ea655b3 the integration-test profile is not used anywhere anymore. remove it. (@michaelsembwever)

• 6073a01 Cleanup classes and tests in the io.cassandrareaper.jmx pacakge (@michaelsembwever)

• e2cf31c Update Dockerfile (@elsmorian)

• 630257d Update cassandra-reaper.yml (@elsmorian)

• cfe3040 Add logic for handling table blacklists (@adejanovski)

• 6b6b785 Add REST endpoint for listing tables from a cluster (@adejanovski)

• 7078584 UI changes for blacklisted tables and other minor improvements (@adejanovski)

• 4c3ece2 Schema changes for blacklisted tables (@adejanovski)

• f342866 Add to checkstyle the ConstantName rule. All static final fields are to be UPPER_CASE. (@michaelsembwever)

• 7dc2316 In cassandra storage override pooling options to allow bigger queue and longer timeout. (@michaelsembwever)

• e42ca22 Add useAddressTranslator and localDC Docker envar support (@joaquincasares)

• ce8e4e2 Fix yaml reference text (@joaquincasares)

• b30652c Add for Docker, which defaults to 20s, based on the code default (@joaquincasares)

• bdb0bd1 Fix for properly handling reboots of Reaper instances with distributed storage (@adejanovski)

• b975c39 rebuilt docs (@rustyrazorblade)

• #218 Docs usage section (#218) (@rustyrazorblade)

• aff6d33 Add useAddressTranslator info in the docs (@adejanovski)

• 26bc722 Add JMX connection test for the whole cluster on init to speed up subsequent connections (@adejanovski)

• 3f4f09e Fix bugs detected by error prone (@adejanovski)

• 354fe59 Ignore any AssertionError trying to releaseLead() when handling a jmx ABORT|ERROR|SESSION_FAILED notification. (@michaelsembwever)

• 2aff661 Increase timeouts for SegmentRunner in SegmentRunnerTest. (@michaelsembwever)

• 61a8ac4 Rename package from com.spotify.reaper to io.cassandrareaper (@adejanovski)

• 8d2846e regen site (@rustyrazorblade)

• d1a5aa6 fixed download links and docker link (@rustyrazorblade)

• fb171d9 Changes requested by Alex (@rustyrazorblade)

• ff6b3ab Format java files to suit checkstyle rules. (@michaelsembwever)

• 9f2b7bb add maven-checkstyle-plugin and a ruleset based off google's codestyle (which seems to be what Spotify was using). (@michaelsembwever)

• f40e87d fixed packages gitignore and page (@rustyrazorblade)

• 92ab864 rebuild website (@rustyrazorblade)

• #203 Improving docs (#203) (@rustyrazorblade)

• 6eb7f9a Fix retry policy NPE errors and disable assertions in packaged installs (@adejanovski)

• eebc3d2 Fix hanging calls when resuming pending repair runs (@adejanovski)

• 2971e07 Add debug logs for potential coordinator selection (@adejanovski)

• e1e436d Add support to run repair on specific nodes or specific datacenter (@adejanovski)

• c483b8d Add tag list inputs for nodes and datacenters in both repair and schedule forms (@adejanovski)

• ae8bce7 Bump version number to 0.8.0 (@adejanovski)

Binaries for the final 0.7.0 fault tolerant version of Reaper.
This version allows running several Reaper instances sharing the same Cassandra backend, and offers full support for multi DC clusters.

Binaries for the 0.7.0 fault tolerant version of Reaper.
This version allows running several Reaper instances sharing the same Cassandra backend, and offers full support for multi DC clusters.

Changelog :

• Fix coordinator node being nulled by abort() in incremental repairs
• Added delete-schedule to the cli (spreaper)
• Allow schedules for both full and incremental repairs on the same keyspace
• Fix ClassCastException when using H2 backend
• Fix SSL encryption communications through JMX
• Fix cluster/rack names with dashes in the UI
• Fix down status not being correctly caught by Reaper when nodes aren't cleanly stopped
• Add docker-compose environment to build packages (jar, deb and rpm)

Changelog :

• Fix for storing run history in schedules with the Postgres/H2 backends

Changelog :

• New data model for the Cassandra backend - huge performance optimizations for this backend
• Fix parallel repair computation for clusters with vnodes
• Minimum wait time between segments is now configurable
• Added an optional query logger for the Cassandra backend

Changelog :

• Add configurable auto-scheduling of repairs for all (non system) keyspaces
• Fix bad segment count display on scheduled repairs
• Add dynamic discovery of seeds for clusters

The JMX Mbeans used to check the number of active and pending repair sessions has changed in Cassandra 2.2, making Reaper unable to check if a repair is already running before running a new segment.

This release fixes the issue by using the right MBean depending on the version of Cassandra.
It also adds a check to prevent running new segments on nodes that are performing validation compaction, and raise the timeout for each segments when running incremental repair (there's one "fake" segment per node in incremental repair, which may take longer than 30mn due to anticompaction).