Create CSV telemetry instances

To create an instance of the CSV provider, build a YAML file (instance.yml in this example) with the definition of the instance:

# CSV Telemetry Provider Instance
provider: CSV File
config:
  address: host1.example.com
  authType: password
  username: akamas
  auth: akamas
  remoteFilePattern: /monitoring/result-*.csv
  componentColumn: COMPONENT
  timestampColumn: TS
  timestampFormat: YYYY-MM-dd'T'HH:mm:ss
metrics:
  - metric: cpu_util
    datasourceMetric: user%

Then you can create the instance for the system using the Akamas CLI:

akamas create telemetry-instance instance.yml system

timestampFormat format

Notice that the week-year format YYYY is compliant with the ISO-8601 specification, but you should replace it with the year-of-era format yyyy if you are specifying a timestampFormat different from the ISO one. For example:

• Correct: yyyy-MM-dd HH:mm:ss

• Wrong: YYYY-MM-dd HH:mm:ss

You can find detailed information on timestamp patterns in the Patterns for Formatting and Parsing section on the DateTimeFormatter (Java Platform SE 8) page.

Configuration options

When you create an instance of the CSV provider, you should specify some configuration information to allow the provider to correctly extract and process metrics from your CSV files.

You can specify configuration information within the config part of the YAML of the instance definition.

Required properties

• address - a URL or IP identifying the address of the host where CSV files reside

• username - the username used when connecting to the host

• authType - the type of authentication to use when connecting to the file host; either password or key

• auth - the authentication credential; either a password or a key according to authType. When using keys, the value can either be the value of the key or the path of the file to import from

• remoteFilePattern - a list of remote files to be imported

Optional properties

• protocol - the protocol to use to retrieve files; either scp or sftp. Default is scp

• fieldSeparator - the character used as a field separator in the CSV files. Default is ,

• componentColumn - the header of the column containing the name of the component. Default is COMPONENT

• timestampColumn - the header of the column containing the timestamp. Default is TS

• timestampFormat - the format of the timestamp (e.g. yyyy-MM-dd HH:mm:ss zzz). Default is YYYY-MM-ddTHH:mm:ss

You should also specify the mapping between the metrics available in your CSV files and those provided by Akamas. This can be done in the metrics section of the telemetry instance configuration. To map a custom metric you should specify at least the following properties:

• metric - the name of a metric in Akamas

• datasourceMetric - the header of a column that contains the metric in the CSV file

The provider ignores any column not present as datasourceMetric in this section.

The sample configuration reported in this section would import the metric cpu_util from CSV files formatted as in the example below:

TS,                   COMPONENT,  user%
2020-04-17T09:46:30,  host,       20
2020-04-17T09:46:35,  host,       23
2020-04-17T09:46:40,  host,       32
2020-04-17T09:46:45,  host,       21

Telemetry instance reference

The following represents the complete configuration reference for the telemetry provider instance.

provider: CSV File             # this is an instance of the CSV provider
config:
  address: host1.example.com   # the address of the host with the CSV files
  port: 22                     # the port used to connect
  authType: password           # the authentication method
  username: akamas             # the username used to connect
  auth: akamas                 # the authentication credential
  protocol: scp                # the protocol used to retrieve the file
  fieldSeparator: ","          # the character used as field separator in the CSV files
  remoteFilePattern: /monitoring/result-*.csv    # the path of the CSV files to import
  componentColumn: COMPONENT                     # the header of the column with component names
  timestampColumn: TS                            # the header of the column with the time stamp
  timestampFormat: YYYY-mm-ddTHH:MM:ss           # the format of the timestamp
metrics:
  - metric: cpu_util                             # the name of the Akamas metric
    datasourceMetric: user%                      # the header of the column with the original metric
    staticLabels:
      mode: user                                 # (optional) additional labels to add to the metric

The following table reports the configuration reference for the config section

Field	Type	Description	Default Value	Restrictions	Required
address	String	The address of the machine where the CSV file resides		A valid URL or IP	Yes
port	Number (integer)	The port to connect to, in order to retrieve the file	22	1≤port≤65536	No
username	String	The username to use in order to connect to the remote machine			Yes
protocol	String	The protocol used to connect to the remote machine: SCP or SFTP	scp	scp sftp	No
authType	String	Specify which method is used to authenticate against the remote machine: password: use the value of the parameter auth as a password key: use the value of the parameter auth as a private key. Supported formats are RSA and DSA		password key	Yes
auth	String	A password or an RSA/DSA key (as YAML multi-line string, keeping new lines)			Yes
remoteFilePattern	String	The path of the remote file(s) to be analyzed. The path can contains GLOB expressio		A list of valid path for linux	Yes
componentColumn	String	The CSV column containing the name of the component. The column's values must match (case sensitive) the name of a component specified in the System	COMPONENT	The column must exists in the CSV file	Yes
timestampColumn	String	The CSV column containing the timestamps of the samples	TS	The column must exists in the CSV file	No
timestampFormat	String	Timestamps' format	YYYY-mm-ddTHH:MM:ss	Must be specified using Java syntax.	No
fieldSeparator	String	Specify the field separator of the CSV	,	, ;	No

The following table reports the configuration reference for the metrics section

Field	Type	Description	Restrictions	Required
metric	String	The name of the metric in Akamas	An existing Akamas metric	Yes
datasourceMetric	String	The name (header) of the column that contains the specific metric	An existing column in the CSV file	Yes
scale	Decimal number	The scale factor to apply when importing the metric
staticLabels	List of key-value pairs	A list of key-value pairs that will be attached to the specific metric sample		No

Use cases

Here you can find common use cases addressed by this provider.

Linux SAR

In this use case, you are going to import some metrics coming from SAR, a popular UNIX tool to monitor system resources. SAR can export CSV files in the following format.

hostname, interval,     timestamp, 		        %user,	%system,      %memory
machine1, 600,		2018-08-07 06:45:01 UTC,	30.01,	20.77,		96.21
machine1, 600,		2018-08-07 06:55:01 UTC,	40.07,	13.00,		84.55
machine1, 600,		2018-08-07 07:05:01 UTC,	5.00,	90.55,		89.23

Note that the metrics are percentages (between 1 and 100), while Akamas accepts percentages as values between 0 and 1, therefore each metric in this configuration has a scale factor of 0.001.

You can import the two CPU metrics and the memory metric from a SAR log using the following telemetry instance configuration.

provider: CSV File
config:
  remoteFilePattern: /csv/sar.csv
  address: 127.0.0.1
  port: 22
  username: user123
  auth: password123
  authType: password
  protocol: scp
  componentColumn: hostname
  timestampColumn: timestamp
  timestampFormat: yyyy-MM-dd HH:mm:ss zzz
metrics:
  - metric: cpu_util
    datasourceMetric: %user
    scale: 0.001
    staticLabels:
      mode: user
  - metric: cpu_util
    datasourceMetric: %system
    scale: 0.001
    staticLabels:
      mode: system
  - metric: mem_util
    scale: 0.001
    datasourceMetric: %memory

Using the configured instance, the CSV File provider will perform the following operations to import the metrics:

1. Retrieve the file "/csv/sar.csv" from the server "127.0.0.1" using the SCP protocol authenticating with the provided password.

2. Use the column hostname to lookup components by name.

3. Use the column timestamp to find the timestamps of the samples (that are expected to be in the format specified by timestampFormat).

4. Collect the metrics (two with the same name, but different labels, and one with a different name):

   • cpu_util: in the CSV file is in the column %user and attach to its samples the label "mode" with value "user".

   • cpu_util: in the CSV file is in the column %system and attach to its samples the label "mode" with value "system".

   • mem_util: in the CSV file is in the column %memory.