6. Create specific dashboards¶

You can create your own Grafana dashboards and Kibana dashboards.

When created, it is recommended to save/export your custom dashboard(s).

6.1. Create specific Grafana dashboards¶

Check the official Grafana documentation and the mfdata_additional_tutorials:Implement custom monitoring and metrics in a plugin tutorial.

6.2. Create specific Kibana dashboards¶

Check the official Kibana documentation and the Kibana - Building your own dashboards tutorial

7. Useful links and Tips¶

7.1. Telegraf error (HTTP 403)¶

Tip

If you do not see any metrics in your Grafana dashboard, check the telegraf.log file in the log directory of the root MFDATA directory.

If you see [agent] Error writing to output [influxdb]: could not write any address and/or net/http: request canceled (Client.Timeout exceeded while awaiting headers), you are probably behind a proxy, and this could happen when MFADMIN and MFDATA are not runinng on the same machine.

You should try to fix this issue by setting the HTTP_PROXY, HTTPS_PROXY and/or NO_PROXY in the /etc/default/telegraf file or the /etc/profile.d/proxy.sh, /etc/profile, e.g. NO_PROXY=”*”

7.2. Querying the InfluxDB database¶

Tip

You may query the InfluxDB database from the InfluxDB’s command line interface influx located in /opt/metwork-mfadmin-master/opt/monitoring/bin/influx.

Check Tools ‘InfluxDB command line interface (CLI/shell)’ menu from the InfluxDB documentation.

The MFADMIN database name is metrics. The default port is 18086.

Examples:

Connect to the influx database

# Connect to the influx database
/opt/metwork-mfadmin-master/opt/metrics/bin/influx -database 'metrics' -host 'localhost' -port '18086'

Connected to http://localhost:18086 version 1.7.2
InfluxDB shell version: 1.7.2
Enter an InfluxQL query
>

List the databases
```
> show databases
```
List the retention policies
```
> show retention policies
```
List the series (the collection of data):
```
> show series
```

List measurements (metrics):

# Query the cpu metrics
> show measurements

Querying measurements (metrics):

# Query the cpu metrics
> select * from cpu

# Query the postgresql metrics
> select * from postgresql

# Query the grib_to_netcdf_metrics metrics
> select * from grib_to_netcdf_metrics

# Query the filecount metrics
> select * from filecount

7.3. Saving your own dashboards¶

Important

When you upgrade MFADMIN Metwork package, the Grafana dashboards and Kibana dashboards are overwritten, so that you may lose the custom dashboards you designed.

So, it’s highly recommended to export the dashboards you want to keep, and the import the dashboards once the Metwork upgrade is done.

In order to export (import) Grafana dashboards, check Grafana documentation.

In order to export (import) Kibana dashboards, check Kibana documentation or use the /opt/metwork-mfadmin/bin/export_kibana_dashboard.py script for exporting and the /opt/metwork-mfadmin/bin/import_kibana_dashboard.py script for importing.

If you build custom dashboards, you would like to be able to reload them automatically when MFADMIN service is started.

In order to do this, you have to store them:

in the /etc/metwork.config.d/mfadmin/grafana_dashboards for Grafana dashboards
in the /etc/metwork.config.d/mfadmin/kibana_dashboards for Kibana dashboards

7.4. Saving the InfluxDB database¶

Important

When you upgrade MFADMIN Metwork package, the InfluxDB database that contains metrics is overwritten.

So, it’s recommended to backup the InfluxDB database, and the restore it once the Metwork upgrade is done.

In order to backup/restore InfluxDB database, check Administration ‘Backing up and restoring’ menu from the InfluxDB documentation.

The name of the MFADMIN Influxdb database is metrics

7.5. Metrics retention (resolution)¶

7.5.1. InfluxDB retention policies and statistics aggregator¶

In InfluxDB databases, data are kept for a limited time according retention policies, i.e., among other things, how long InfluxDb keeps the data (duration).

In MFADMIN, Telegraf is configured for emitting new aggregate metrics, such as min, max, mean, sum (basic statistics aggregator plugin). This aggregator is configured with a period. The period is the size of the window of metrics that each aggregate represents. In other words, the emitted aggregate metric will be the aggregated value of the past period seconds.

MFADMIN supplies the following retention policies:

fullres : data are kept for 24 hours, statistics aggregation period is 10 seconds
medres : data are kept for 7 days, statistics aggregation period is 60 seconds
lowres : data are kept for 52 weeks, statistics aggregation period is 1 hour

In the Grafana dashboards, you can display time-series metrics with the suitable resolution.

7.5.2. Elasticsearch retention¶

ElasticSearch no longer supports retention period. So, the is no default data retention period.

To avoid having an ElasticSearch database growing indefinitely, a MFADMIN elasticsearch.clean python script is running at regular intervals in order to delete ElasticSearch indices based on age.

The age (lifetime) is configured in the indices_lifetime parameter [elasticsearch] section of the MFADMIN configuration file file, the default value is 5 days, e.g.:

[elasticsearch]

...

# indices starting with "filebeat-" are destroyed after {indices_lifetime} days
# -1 => no cleaning
# indices_lifetime = 5

7.6. Exporting logs¶

This section applies to all MetWork modules.

Logs are exported in a separate json file. This could be useful if you need monitoring with external tool.

In order to save logs in a separate file, you have to check/set the json_file and json_minimal_level parameters in the [log] section of the config/config.ini file stored in the root directory of the MetWork module user.

For further details, refer to MetWork plugins configuration.

7.7. Circus hooks¶

Circus provides hooks used to trigger actions upon watcher events. These hooks are configured in the circus.ini configuration file.

MFADMIN circus configuration, like the other MetWork modules (MFBASE, MFDATA, MFSERV, …), implements circus hooks for circus watchers (e.g. before_start, after_start hooks).

Check the circus.ini configuration file in the tmp/config_auto/ in the root directory of the mfadmin user (or other MetWork modules, e.g. mfbase, mfdata, mfserv).

...
[watcher:influxdb]
...
hooks.before_start=circus_hooks.before_start_shell
hooks.after_start=circus_hooks.after_start_shell
...

In the above example, a shell script is executed before influxdb (hooks.before_start=circus_hooks.before_start_shell) starts and after influxdb is started (hooks.before_start=circus_hooks.before_start_shell).

The shell script name is {hookname}{whatcher_name}, e.g:

before_start_influxdb
after_start_influxdb

You will find the hooks script in the bin directory of the installed MetWork module, e.g. /opt/metwork-mfadmin-master/bin/