- Update to v1.20.0 (go.d.plugin v0.15.0)
Breaking Changes:
* Removed deprecated bash collectors apache, cpu_apps, cpufreq,
exim, hddtemp, load_average, mem_apps, mysql, nginx, phpfpm,
postfix, squid, tomcat. If you were still using one of these
collectors with custom configurations, you can find the new
collector that replaces it in the supported collectors list.
Improvements:
* Host labels:
+ Added support for host labels
+ Improved the monitored system information detection.
Added CPU freq & cores, RAM and disk space.
+ Started distinguishing the monitored system's (host)
OS/Kernel etc. from those of the docker container's
+ Started creating host labels from collected system info
+ Started passing labels and container environment variables
via the streaming protocol
+ Started sending host labels via exporting connectors
+ Added label support to alarm definitions and started
recording them in alarm logs
+ Added support for host labels to the API responses
+ Added configurable host labels to netdata.conf
+ Added kubernetes labels
* New collectors:
+ eBPF kernel collector
+ CockroachDB
+ squidlog: squid access log parser
* Collector improvements:
+ apps.plugin: Created dns group, improved database group,
improved ceph & samba groups
+ varnish: Added SMF metrics (cache on disk)
+ phpfpm: Fixed per process chart titles and readme
+ python.d: Formatted the code in all modules
+ node.d/snmp: - Added snmpv3 support, formatted the code in
snmp.node.js
+ cgroups: Improved LXC monitoring by filtering out irrelevant
LXC cgroups
+ litespeed: Added support for different .rtreport format
+ proc.plugin: Added pressure stall information
+ sensors: Improved collection logic
+ proc: Started monitoring network interface speed, duplex,
operstate
+ smartd_log: Fixed the setting in the reallocated sectors
count, by setting ATTR5 chart algorithm to absolute
+ nvidia-smi: Allow executing nvidia-smi in normal instead of
loop mode
+ wmi: collect logon metrics, collect logical_disk disk
latency metrics
+ weblog: handle MKCOL, PROPFIND, MOVE, SEARCH http request
methods
+ scaleio: storage pools and sdcs metrics
* Exporting engine:
+ Implemented the main flow for the Exporting Engine
* Streaming:
+ Add versioning to the streaming protocol
* Installation/Packages:
+ Improved the systemd service files, by removing unecessary
ExecStartPre lines and moving global options to netdata.conf
* Privacy:
+ Added support for opting out of telemetry via the
DO_NOT_TRACK environment variable (telemetry is disabled by
default on openSUSE)
* Other:
+ Preparations for the next netdata cloud release. Added custom
libmosquitto, netdata-cli and other prerequisites.
+ Applied linter fixes in shell scripts
+ Started supporting -fno-common in CFLAGS
+ Completely removed the unbound python collector (dead code)
+ Added possibility to change badges' text font color
+ Small updates to sample multi-host dashboard, dash.html
+ Added missing quoting in shell scripts
+ Bump handlebars from 4.2.0 to 4.5.3
+ Reduce log level for uv_thread_set_name_np from error to info
+ Set standard name to non-libnetdata threads (libuv, pthread)
* Bug fixes:
+ Fixed problems reported by Coverity for eBPF collector plugin
+ Fixed invalid literal for float\(\): NN.NNt error in the
elasticsearch python plugin, by adding terabyte unit parsing
+ Fixed timeout failing in docker containers which broke some
python.d collectors
+ Fixed problem with the httpcheck python collector not being
able to check URLs with the POST method, by adding body to
the URLService. Also record the new options in httpcheck.conf
+ Fixed dbengine repeated global flushing errors and collectors
being blocked, by dropping dirty dbengine pages if the disk
cannot keep up
+ Fixed issue with alarm notifications occasionally ignoring
the configured severity filter when the ROLE was set to root
+ Fixed Netlink Connection Tracker charts in the nfacct plugin
+ Fixed metric values sent via remote write to Prometheus
backends, when using average/sum
+ Fixed unclosed brackets in softnet alarm
+ Fixed SEGFAULT when localhost initialization failed
+ Reduced the number of broken pipe error log entries,
after a SIGKILL
+ Fixed Coverity errors by restoring support for protobuf 3.0
+ Fixed inability to disable Prometheus remote API
+ Fixed SEGFAULT from the cpuidle plugin
+ Fixed samba collector not working, due to inability to run
sudo
+ Fixed invalid css/js resource errors when URL for slave node
has no final / on streaming master
+ Fixed keys_redis chart in the redis collector, by populating
keys at runtime
+ Fixed UrlService bytes decoding and logger unicode encoding
in the python.d plugin
+ Fixed a warning in the prometheus remote write backend
+ Fixed not detecting more than one adapter in the hpssa
collector
+ Fixed race conditions in dbengine
+ Fixed race condition with the dbenging page cache descriptors
+ Fixed dbengine dirty page flushing warning
+ Fixed missing parenthesis on alarm softnet.conf
+ Fixed "Master thread EXPORTING takes too long to exit.
Giving up" error, by cleaning up the main exporting engine
thread on exit
+ Fixed rabbitmq error "update() unhandled exception: invalid
literal for int() with base 10"
+ Fixed some LGTM alerts
+ Fixed valgrind errors
+ Fixed monit collector LGTM warnings
+ Fixed the following go.d.plugin collector issues:
. mysql: panic in Cleanup (#326)
. unbound: gather metrics via unix socket (#319)
. logstash: pipelines chart (#317)
. unbound: configuration file parsing.
. Support include mechanism. (#298)
. logstash: pipelines metrics parsing (#293)
. phpfpm: processes metrics parsing (#297)