All notable changes to this project will be documented in this file. This project adheres to Semantic Versioning.
- Dropped Support for Python 2.x, <=3.4.x
- A namespace prefix can be added to all metrics. #434
- An optional status monitoring API, serving JSON over HTTP #422
- Command-line options can be configured using an optional
/etc/tcollector.json
configuration file. #433 - Boolean values True, true, False, and false, are now converted to 1,1,0 and 0 respectively. No PR #, I merged it to master directly...sorry....
- collectors/0/nfsstat: correct list of nfs client names #374
- Adding two TCP stats to netstat.py #419
- Add info on the cpu sets to procstats #432
- Ignore kubernetes mounts in dfstat.py #418
- Adding ncat usage option to opentsdb.sh #420
- updated HBase ports for modern versions of HBase (0.99+ @ 2014/2015) #394
- iostat - Generate write_bytes and read_bytes metrics enhancement #326
- allow dates past Sept 2020 #405
- additonial Python compatibility #396 #411 #218
- Added mysqlclient module #383
- Collector for MapR Hadoop node metrics #281
- New Collectors for Hadoop Yarn Resource Manager and Node Manager JMX API Stats #400
- netfilter stats #354
- Added postgresql_replication collector. #323
- Adding buddying memory fragmentation and slab info stats for tcollector #318
- Added mountstats collector #322
- tcollector daemon hangs and becomes unresponsive. #378
- If response code from OpenTSDB is 400, don't retry sending since this means we sent bad data. #436
- Small data collection validation refinements #431
- fix elasticsearch sending booleans to opentsdb #429
- Fix order of checking if value is bool #428
- add flush to zookeeper collector agent #427
- correctly dedup timestamps in milleseconds #440
- fix proc status #425
- Hbase metric lost due to coding problems In python2 #439
- hadoop_http.py - Fixed unicode issue #437
- hbase regionserver collector "Error splitting numRegions" #396
- added fix for #398 to flush metrics after each emit_metric #399
- Fix for multithreaded HAProxy (since HAProxy 1.8) #404
- Fix memory leak around timestamp precision adjust #408
- Check counter names exist before referencing. #377 #376
- docker.py - Pulls metrics from a local Docker instance, tries /var/run/docker.sock, then localhost API
- pxc-collector.py - Added Percona XtraDB Cluster Collector #301
- mongo3.py - Added MongoDB 3 Collector #302
- startstop - Fixed issue where host was still required #291
- tcollector.py - Fixed default pid location #299
- tcollector.py - Few bugs related to new configuration [#305(#305) #306 #307
- tcollector.py - Fixed issues with default cdir and classpath on Debian #308
- tcollector.py - Fixed issue with last_datapoint for longer running collectors #309
- Improved ZFS Iostat collector #294
- Avoid picking new connction with just one host #295
- Improved CPI pctusage #298
- Fixed CGROUP path for EL7 #304
- rtt.py - Pings a configured set of hosts and records the result #183
- aws_cloudwatch.py - Collects metrics from AWS Cloudwatch #246
- tcp_bridge.py - Listens on a TCP port for metrics to relay to configured hosts #252
- ntp.py - Gathers offset of clock from NTP #265
- mapr_metrics.py - Gathers MAPR metrics for Hadoop nodes #196
- tcollector.py - Gathers metrics on CPU and Memory usage of TCollector and the collectors #276
- Support for HTTP API #225
- Support for MS precision #230 #234
- Support TSD_HOSTS comma delimited host list rather than TSD_HOST/TSD_PORT #237
- RPM Packages now have base TCollector package, Collectors Package and an EOS specific package #244
- Improved FreeBSD compatiblity #256 #257 #258 #263
- TCollector and all collectors now using '/usr/bin/env python' rather than '/usr/bin/python' #263
- Elasticsearch Collector now allows multiple+configurable targets #207
- TCollector now reads default values from the configuration #287
- The RPM no longer starts or enables TCollector on installation
- Enable RPM subpackages. Allows you to create 3 packages
- A base package with the just the tcollector
- A collectors subpackage with just collectors
- An eos subpackage with EOS-only collectors (depends on EosSDK)
- %{rootdir} was hardcoded instead of using the pwd passed to %{_srcrpmdir}.
- The sub-packages were missing a dependency on the main package.
- This is the current release, so the CHANGELOG is from here forward.