cc-metric-collector/docs/configuration.md
Thomas Gruber 3f76947f54
Merge latest developments into main (#67)
* Update configuration.md

Add an additional receiver to have better alignment of components

* Change default GpfsCollector command to `mmpmon` (#53)

* Set default cmd to 'mmpmon'

* Reuse looked up path

* Cast const to string

* Just download LIKWID to get the headers (#54)

* Just download LIKWID to get the headers

* Remove perl-Data-Dumper from BuildRequires, only required by LIKWID build

* Add HttpReceiver as counterpart to the HttpSink (#49)

* Use GBytes as unit for large memory numbers

* Make maxForward configurable, save old name in meta in rename metrics and make the hostname tag key configurable

* Single release action (#55)

Building all RPMs and releasing in a single workflow

* Makefile target to build binary-only Debian packages (#61)

* Add 'install' and 'DEB' make targets to build binary-only Debian packages

* Add control file for DEB builds

* Use a single line for bash loop in make clean

* Add config options for retry intervals of InfluxDB clients (#59)

* Refactoring of LikwidCollector and metric units (#62)

* Reduce complexity of LikwidCollector and allow metric units

* Add unit to LikwidCollector docu and fix some typos

* Make library path configurable

* Use old metric name in Ganglia if rename has happened in the router (#60)

* Use old metric name if rename has happened in the router

* Also check for Ganglia renames for the oldname

* Derived metrics (#57)

* Add time-based derivatived (e.g. bandwidth) to some collectors

* Add documentation

* Add comments

* Fix: Only compute rates with a valid previous state

* Only compute rates with a valid previous state

* Define const values for net/dev fields

* Set default config values

* Add comments

* Refactor: Consolidate data structures

* Refactor: Consolidate data structures

* Refactor: Avoid struct deep copy

* Refactor: Avoid redundant tag maps

* Refactor: Use int64 type for absolut values

Co-authored-by: Holger Obermaier <40787752+ho-ob@users.noreply.github.com>

* Simplified iota usage

* Move unit tag to meta data tags

* Derived metrics (#65)

* Add time-based derivatived (e.g. bandwidth) to some collectors

* Add documentation

* Add comments

* Fix: Only compute rates with a valid previous state

* Only compute rates with a valid previous state

* Define const values for net/dev fields

* Set default config values

* Add comments

* Refactor: Consolidate data structures

* Refactor: Consolidate data structures

* Refactor: Avoid struct deep copy

* Refactor: Avoid redundant tag maps

* Refactor: Use int64 type for absolut values

* Update LustreCollector

Co-authored-by: Holger Obermaier <40787752+ho-ob@users.noreply.github.com>

* Meta to tags list and map for sinks (#63)

* Change ccMetric->Influx functions

* Use a meta_as_tags string list in config but create a lookup map afterwards

* Add meta as tag logic to sampleSink

* Fix staticcheck warnings (#66)

Co-authored-by: Holger Obermaier <40787752+ho-ob@users.noreply.github.com>
2022-03-15 16:41:11 +01:00

6.9 KiB

Configuring the CC metric collector

The configuration of the CC metric collector consists of five configuration files: one global file and four component related files.

Global configuration

The global file contains the paths to the other four files and some global options.

{
  "sinks": "sinks.json",
  "collectors" : "collectors.json",
  "receivers" : "receivers.json",
  "router" : "router.json",
  "interval": 10,
  "duration": 1
}

Be aware that the paths are relative to the execution folder of the cc-metric-collector binary, so it is recommended to use absolute paths.

Component configuration

The others are mainly list of of subcomponents: the collectors, the receivers, the router and the sinks. Their role is best shown in a picture:

flowchart LR

  subgraph col ["Collectors"]
  direction TB
  cpustat["cpustat"]
  memstat["memstat"]
  tempstat["tempstat"]
  misc["..."]
  end
  
  subgraph Receivers ["Receivers"]
  direction TB
  nats["NATS"]
  httprecv["HTTP"]
  miscrecv[...]
  end

  subgraph calc["Aggregator"]
  direction LR
  cache["Cache"]
  agg["Calculator"]
  end

  subgraph sinks ["Sinks"]
  direction RL
  influx["InfluxDB"]
  ganglia["Ganglia"]
  logger["Logfile"]
  miscsink["..."]
  end

  cpustat --> CollectorManager["CollectorManager"]
  memstat --> CollectorManager
  tempstat --> CollectorManager
  misc --> CollectorManager

  nats  --> ReceiverManager["ReceiverManager"]
  httprecv --> ReceiverManager
  miscrecv --> ReceiverManager

  CollectorManager --> newrouter["Router"]
  ReceiverManager -.-> newrouter
  calc -.-> newrouter
  newrouter --> SinkManager["SinkManager"]
  newrouter -.-> calc

  SinkManager --> influx
  SinkManager --> ganglia
  SinkManager --> logger
  SinkManager --> miscsink


There are four parts:

  • The collectors read data from files, execute commands and call dynamically loaded library function and send it to the router
  • The router can process metrics by cacheing and evaluating functions and conditions on them
  • The sinks send the metrics to storage backends
  • The receivers can be used to receive metrics from other collectors and forward them to the router. They can be used to create a tree-like structure of collectors.

(A maybe better differentiation between collectors and receivers is that the collectors are called periodically while the receivers have their own logic and submit metrics at any time)

Collectors configuration file

The collectors configuration file tells which metrics should be queried from the system. The metric gathering is logically grouped in so called 'Collectors'. So there are Collectors to read CPU, memory or filesystem statistics. The collectors configuration file is a list of these collectors with collector-specific configurations:

{
  "cpustat" : {},
  "diskstat": {
    "exclude_metrics": [
      "disk_total"
    ]
  }
}

The first one is the CPU statistics collector without any collector-specific setting. The second one enables disk mount statistics but excludes the metric disk_total.

All names and possible collector-specific configuration options can be found here.

Some collectors might dynamically load shared libraries. In order to enable these collectors, make sure that the shared library path is part of the LD_LIBRARY_PATH environment variable.

Sinks configuration file

The sinks define the output/sending of metrics. The metrics can be forwarded to multiple sinks, even to sinks of the same type. The sinks configuration file is a list of these sinks, each with an individual name.

{
  "myinflux" : {
    "type" : "influxasync",
    "host": "localhost",
    "port": "8086",
    "organization" : "testorga",
    "database" : "testbucket",
    "password" : "<my secret JWT>"
  },
  "companyinflux" : {
    "type" : "influxasync",
    "host": "companyhost",
    "port": "8086",
    "organization" : "company",
    "database" : "main",
    "password" : "<company's secret JWT>"
  }
}

The above example configuration file defines two sink, both ot type influxasync. They are differentiated internally by the names: myinflux and companyinflux.

All types and possible sink-specific configuration options can be found here.

Some sinks might dynamically load shared libraries. In order to enable these sinks, make sure that the shared library path is part of the LD_LIBRARY_PATH environment variable.

Router configuration file

The collectors and the sinks are connected through the router. The router forwards the metrics to the sinks but enables some data processing. A common example is to tag all passing metrics like adding cluster=mycluster. But also aggregations like "take the average of all 'ipc' metrics" (ipc -> Instructions Per Cycle). Since the configurations of these aggregations can be quite complicated, we refer to the router's README.

A simple router configuration file to start with looks like this:

{
    "add_tags" : [
        {
            "key" : "cluster",
            "value" : "mycluster",
            "if" : "*"
        }
    ],
    "interval_timestamp" : false,
    "num_cache_intervals" : 0
}

With the add_tags section, we tell to attach the cluster=mycluster tag to each (* metric). The interval_timestamp tell the router to not touch the timestamp of metrics. It is possible to send all metrics within an interval with a common time stamp to avoid later alignment issues. The num_cache_intervals diables the cache completely. The cache is only required if you want to do complex metric aggregations.

All configuration options can be found here.

Receivers configuration file

The receivers are a special feature of the CC Metric Collector to enable simpler integration into exising setups. While collectors query data from the local system, the receivers commonly get data from other systems through some network technology like HTTP or NATS. The idea is keep the current setup but send it to a CC Metric Collector which forwards it to the the destination system (if a sink exists for it). For most setups, the receivers are not required and an the receiver config file should contain only an empty JSON map ({}).

{
  "nats_rack0": {
    "type": "nats",
    "address" : "nats-server.example.org",
    "port" : "4222",
    "subject" : "rack0",
  },
  "nats_rack1": {
    "type": "nats",
    "address" : "nats-server.example.org",
    "port" : "4222",
    "subject" : "rack1",
  }
}

This example configuration creates two receivers with the names nats_rack0 and nats_rack1. While one subscribes to metrics published with the rack0 subject, the other one subscribes to the rack0 subject. The NATS server is the same as it manages all subjects in a subnet. (As example, the router could add tags rack=0 and rack=1 respectively to the received metrics.)

All types and possible receiver-specific configuration options can be found here.