All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog and this project adheres to Semantic Versioning.
- Add import of distutils.util to settings.py fixing missing module error
0.14.14 - 2019-06-12
- Freeze cassandra-driver to version to 3.17.1
- Constrain elasticsearch-dsl to versions <7.0.0,>=6.0.0
- Creation rate of metadata configurable
- Ability to disable metadata creation at runtime
0.14.13 - 2019-04-12
- Version binding of dependencies to stay python 2.x compatible
- Freeze gourde version to 0.2.1
0.14.12 - 2019-04-01
- Add tracing on the graphite plugin interface
- Align the way to update metadata in Elasticsearch with Cassandra
- Setting BG_CASSANDRA_READ_ON_SAMPLING_RATE to 0 skips updating the read_on
- Do not update read_on if the metric is not present
0.14.11 - 2019-02-27
- Add new configuration keys to Elasticsearch driver
0.14.10 - 2019-02-19
- Wait metadata to be read before touching them
0.14.9 - 2019-02-14
- Pass correct time argument to carbonlink merge function
- Adding docker-compose setup to test biggraphite
- Fixed accessor cache on python 3.x
- Fixed bug in ES metrics_matching_glob
- Fixed data read/write consistency level in cassandra driver
0.14.8 - 2019-01-04
- crash when opencensus is not installed
0.14.7 - 2018-12-27
- Tracing shows more relevant data in spans (metric names/globs if available)
- Middleware is provided to improve tracing:
- more options to configure when to trace
- more data in spans (the executed query and the format)
0.14.6 - 2018-12-18
- reduced number of metadata objects handled by biggraphite
- some bgutils commands were printing python objects instead of string representation
0.14.5 - 2018-12-13
- adapt biggraphite to use the lastest version of prometheus client
0.14.4 - 2018-12-13
- monitoring: some Cassandra counters were stuck at 0
- Cassandra backend returned empty results in traced environment
- specifing the same data driver as metadata driver lead to runtime errors
- diskcache not used with Cassandra
0.14.3 - 2018-11-26
- fix list index out of range in _get_metrics when the metric doesn't exists
- support tracing of biggraphite
- expose metrics from the elasticsearch driver
0.14.2 - 2018-10-18
- CVE-2017-12629 on graphiteindex (upgrade to Lucene 7.1.0)
0.14.1 - 2018-10-15
- improved observability on metric creation
- performance of Elasticsearch driver
- isolation between metadata and data in Cassandra driver
- metric name UTF-8 support in Elasticsearch driver
- ability to disable BigGraphite cache
- irrelevant metrics on
bg_cassandra_updated_on_latency_seconds
andbg_cassandra_read_on_latency_seconds
for Cassandra driver
0.13.10 - 2018-10-08
- cassandra accessor observability: add metrics on query usage
0.13.9 - 2018-08-28
- updated_on behavior
- bgutil stats supports metric prefix
0.13.8 - 2018-08-27
- cassandra: don't loose points if we have many restarts
- Fix python 3.6 compatibility
- bgutil stats: add carbon support
- Add flask_cors support
0.13.7 - 2018-08-23
- elasticsearch: env variables are now working properly
- elasticsearch: timestamps are now updated on a background thread
0.13.6 - 2018-08-22
- elasticsearch: fix read_on bug
- bgutil web: add /maintainance page
0.13.5 - 2018-08-21
- Support clean/stats/repair in ES
- bgutil web: fix and finish async commands
0.13.4 - 2018-08-16
- Don't install ipython as requirement.
0.13.3 - 2018-08-14
- Configurable es schema
- bgutil web: async cmd
- bgutil web: capture stderr/stdout
- ES import script (schema + doc + remove directory + doc type)
- Wrong C* metric metadata parsing
0.13.2 - 2018-08-03
- Configurable es schema (reverted while waiting for a fix)
- Accessor clear method (only used by tests)
0.13.1 - 2018-08-02
- Use provided time range to glob search metrics and directories (Elasticsearch only)
- Metadata update lost creation date in Elasticsearch backend
- Ignore update conflicts on Elasticsearch (HTTP 409)
0.13.0 - 2018-23-07
- Elasticsearch driver (supports metadata only), see ELASTICSEARCH_DESIGN.md
- Ability to declare distinct data and metadata drivers: use
--data_driver
and--metadata-driver
instead of--driver
(--driver
option is still supported)
0.12.0 - 2018-25-06
- bgutil clean new flag: --corrupted-metrics
- Add non-zero timeout to event.wait()
- Make sure we always create missing directories
- Non escaped char in the Lucene filter string
- Fix disk cache and add unitests
0.11.0 - 2018-03-13
- Metrics for metadata cache and cassandra operations.
- Ability to configure SYNC_EVERY_N_WRITES in carbon
- Shuffle Cassandra replicas
- Connection leak when Cassandra connection fails
0.10.7 - 2018-02-15
- Added bgutil graphite_web to debug the graphite web plugin.
- Reduce calls to biggraphite_metadata.metrics_metadata when reading points.
0.10.6 - 2018-02-15
- New options for bg-replay-traffic
- metrics_metadata is now less subject to corruptions
0.10.5 - 2018-02-01
- Cache issues with globs
0.10.4 - 2018-01-31
- Experimental Lucene index (with BG_CASSANDRA_USE_LUCENE).
0.10.3 - 2018-01-15
- Fixed wrong results when using performance improvements for 1.1.x
- Improved the way readers and finders were imported.
0.10.2 - 2018-01-08
- Fixed performance regression with 1.1.0
- Fixed some bugs caused by the Python 3 support
0.10.0 - 2017-12-20
- Support for Graphite 1.1.0
- Support for Python 3
- Add CSV support to bgutil read
- Bug where leaves_only would be ignored when caching values
- Queries returning no results would no be cached correctly
- Add locking in metadata_cache
- Fixed benchmarks
0.9.2 - 2017-12-04
- Fix compatibility with Graphite 1.1.0
0.9.1 - 2017-12-04
- Version in setup.py
- Experiments with Lucence and ES
0.9.0 - 2017-11-21
- The previously non working
--(meta|data)_consistency*
flags have been prefixed bycassandra
.
- Cassandra background operations (clean, repair) are now more resilient to timeouts
- Fixed Cassandra consistency flags.
0.8.11 - 2017-11-08
bgutil stats
: Skip corrupt metricsbgutil write
: fix command and add unit test- tools: Fix columnspec for time_start_ms.
bgutil syncdb
: add a way to create datapoint tables
0.8.10 - 2017-10-12
-
Renamed
created_at
->created_on
: Before updating, run:ALTER TABLE biggraphite_metadata.metrics_metadata ADD created_on timeuuid; CREATE CUSTOM INDEX metrics_metadata_created_on_idx ON biggraphite_metadata.metrics_metadata (created_on) USING 'org.apache.cassandra.index.sasi.SASIIndex' WITH OPTIONS = {'mode': 'SPARSE'};
After updating, run:
DROP CUSTOM INDEX metrics_metadata_created_at_idx; ALTER TABLE biggraphite_metadata.metrics_metadata DROP created_at;
- Carbon: make sure we touch metrics on all code path
bgutil delete
to delete metrics
0.8.9 - 2017-09-26
- Only update metadata timestamps when writing
- New created_at field in metrics_metadata
- Some work for the new GraphiteIndex, not ready yet.
0.8.8 - 2017-07-11
- Bug in cleanup code (would remove intermediate directories)
0.8.7 - 2017-07-07
- Infinite caching for metadata.
0.8.6 - 2017-06-15
- Infinite caching for metadata.
0.8.5 - 2017-06-15
- Fixed the time info of empty timeseries
- Fix error handling in repair()/clean()
- Added bgutil stats
- Better caching for find()
- Use new Graphite-Web API for FindQuery
- Make read_on/updated_on more configurable
0.8.4 - 2017-05-15
- Better logging
- Add
bgutil copy
- Better exception in carbon.py when metric can't be found
0.8.3 - 2017-04-25
- Nothing, just an issue with pypi uploads.
0.8.2 - 2017-04-25
- Fix ipython and cassandra versions
0.8.1 - 2017-04-05
- Better handling of Cassandra exceptions
- Better handling of carbonlink
- Fix metric reporting interval
0.8.0 - 2017-03-10
- Configurable consistency settings (#242)
- [BREAKING] New column (read_on) in metric_metadata for adding statistics on read metrics (#107) need to run the following cql to migrate theschema
ALTER TABLE biggraphite_metadata.metrics_metadata
ADD read_on timeuuid;
- Added a tool to replay graphite traffic
- Support Carbon Link protocol
- Fix multi-cluster support
- Be more robust when metadata is missing
- Added some randomness to cache expiration
- Prometheus support and additional metrics
- Add support for multiple replicas/writers
- Microbenchmarks
- Asynchronous creation of metrics in carbon
- Better bg-import-whisper / bg-generate-sstables + helper scripts
- Multiple bugs with the disk cache
- Multiple clean/repair bugs
- [Breaking] Cassandra schema as been changed to separate stage0 and aggregated metrics. There is no upgrade procedure as this is a pre-release.
- [Breaking]
bgutil syncdb
needs to be run during initial install
0.6.0 - 2016-11-25
- Initial release with this CHANGELOG file
- We are going to do releases from now on