Traffic Stats Administration
Traffic Stats consists of three separate components: Traffic Stats, InfluxDB, and Grafana. See below for information on installing and configuring each component as well as configuring the integration between the three and Traffic Ops.
Installation
Installing Traffic Stats
See the downloads page for Traffic Control to get the latest release.
Follow the instructions in Building Traffic Control to generate an RPM.
Copy the RPM to your server
Install the generated Traffic Stats RPM with yum(8) or rpm(8)
Installing InfluxDB
In order to store Traffic Stats data you will need to install InfluxDB. While not required, it is recommended to use some sort of high availability option like Influx enterprise, InfluxDB Relay, or another high availability option.
Installing Grafana
Grafana is used to display Traffic Stats/InfluxDB data in Traffic Ops. Grafana is typically run on the same server as Traffic Stats but this is not a requirement. Grafana can be installed on any server that can access InfluxDB and can be accessed by Traffic Ops. Documentation on installing Grafana can be found on the Grafana website.
Configuration
Configuring Traffic Stats
Traffic Stats’s configuration file can be found in /opt/traffic_stats/conf/traffic_stats.cfg
. It is a JSON-format file with the following properties.
- toUser:
The username of the user as whom to connect to Traffic Ops
- toPasswd:
The password to use when authenticating with Traffic Ops
- toUrl:
The URL of the Traffic Ops server used by Traffic Stats
- toRequestTimeout:
The time, in seconds, before a client request to Traffic Ops is canceled. Defaults to 10 if no value provided
- influxUser:
Optionally specify the user to use when connecting to InfluxDB (if InfluxDB is not configured to require authentication, has no effect)
- influxPassword:
Optionally specify the password to use when connecting to InfluxDB (if InfluxDB is not configured to require authentication, has no effect)
- pollingInterval:
The interval in seconds for which Traffic Monitor will wait between polling for stats and storing them in InfluxDB
- statusToMon:
The name of the Status of the Traffic Monitors to poll (e.g.
ONLINE
orOFFLINE
)- seelogConfig:
Optionally specify the absolute path to a seelog configuration file. Has no effect if the
logs
property is present.Deprecated since version ATCv6.1: This will be removed in the future, configurations should be migrated to the new
logs
property.- logs:
This property is an object containing keys that specify locations for different log streams.
New in version ATCv6.1.
- error:
The location of error-level logs. If omitted,
null
, an empty string (""
), or the special path/dev/null
(even on Windows), error-level logs will not be emitted. The special values “stderr” and “stdout” cause logging to use STDOUT or STDERR, respectively[1].- warning:
The location of warning-level logs. If omitted,
null
, an empty string (""
), or the special path/dev/null
(even on Windows), warning-level logs will not be emitted. The special values “stderr” and “stdout” cause logging to use STDOUT or STDERR, respectively[1].- info:
The location of info-level logs. If omitted,
null
, an empty string (""
), or the special path/dev/null
(even on Windows), info-level logs will not be emitted. The special values “stderr” and “stdout” cause logging to use STDOUT or STDERR, respectively[1].- debug:
The location of debug-level logs. If omitted,
null
, an empty string (""
), or the special path/dev/null
(even on Windows), debug-level logs will not be emitted. The special values “stderr” and “stdout” cause logging to use STDOUT or STDERR, respectively[1].- event:
The location of event-level logs. If omitted,
null
, an empty string (""
), or the special path/dev/null
(even on Windows), event-level logs will not be emitted. The special values “stderr” and “stdout” cause logging to use STDOUT or STDERR, respectively[1].Note
At the time of this writing, Traffic Stats does not make use of the “event” log level.
- dailySummaryPollingInterval:
The interval, in seconds, on which Traffic Stats checks to see if daily stats need to be computed and stored.
- cacheRetentionPolicy:
The default retention policy for cache stats
- dsRetentionPolicy:
The default retention policy for Delivery Service statistics
- dailySummaryRetentionPolicy:
The retention policy to be used for the daily statistics
- influxUrls:
An array of InfluxDB hosts for Traffic Stats to write stats to.
Configuring InfluxDB
As mentioned above, it is recommended that InfluxDB be running in some sort of high availability configuration. There are several ways to achieve high availability so it is best to consult the high availability options on the InfuxDB website.
Once InfluxDB is installed and configured, databases and retention policies need to be created. Traffic Stats writes to three different databases: cache_stats, deliveryservice_stats, and daily_stats. More information about the databases and what data is stored in each can be found in the Traffic Stats Overview.
To easily create databases, retention policies, and continuous queries, run create_ts_databases from the /opt/traffic_stats/influxdb_tools
directory on your Traffic Stats server. See the InfluxDB Tools section for more information.
Configuring Grafana
In Traffic Portal the
menu item can be configured to display Grafana graphs using InfluxDB data (when not configured, this menu item will not appear). In order for this to work correctly, you will need two things:A Parameter with the graph URL (more information below)
The graphs created in Grafana. See below for how to create some simple graphs in Grafana. These instructions assume that InfluxDB has been configured and that data has been written to it. If this is not true, you will not see any graphs.
To create a graph in Grafana, you can follow these basic steps:
Login to Grafana as an administrative user
Click on
Enter the necessary information to configure your data source
Click on
at the bottomClick on
Where it says No Title (click here) click and choose edit
Choose your data source at the bottom
You can have Grafana help you create a query, or you can create your own.
SELECT sum(value)*1000 FROM "monthly"."bandwidth.cdn.1min" GROUP BY time(60s), cdn;
Once you have the graph the way you want it, click the Save Dashboard button at the top
You should now have a new saved graph
In order for Traffic Portal users to see Grafana graphs, Grafana will need to allow anonymous access. Information on how to configure anonymous access can be found on the configuration page of the Grafana Website.
Traffic Portal uses custom dashboards to display information about individual Delivery Services or Cache Groups. In order for the custom graphs to display correctly, the Javascript files in traffic_stats/grafana/
need to be in the /usr/share/grafana/public/dashboards/
directory on the Grafana server. If your Grafana server is the same as your Traffic Stats server the RPM install process will take care of putting the files in place. If your Grafana server is different from your Traffic Stats server, you will need to manually copy the files to the correct directory.
See also
More information on custom scripted graphs can be found in the scripted dashboards section of the Grafana documentation.
Configuring Traffic Portal for Traffic Stats
The InfluxDB servers need to be added to Traffic Portal with a Profile that has the Type InfluxDB. Make sure to use port 8086 in the configuration.
The traffic stats server should be added to Traffic Ops with a Profile that has the Type TRAFFIC_STATS.
Parameters for which stats will be collected are added with the release, but any changes can be made via Parameters that are assigned to the Traffic Stats Profile.
Configuring Traffic Portal to use Grafana Dashboards
To configure Traffic Portal to use Grafana Dashboards, you need to enter the following Parameters and assign them to the special GLOBAL Profile. This assumes you followed instructions in the Installation, Configuring Traffic Stats, Configuring InfluxDB, and Configuring Grafana sections.
parameter name |
parameter value |
---|---|
all_graph_url |
|
cachegroup_graph_url |
|
deliveryservice_graph_url |
|
server_graph_url |
|
visual_status_panel_1 |
|
visual_status_panel_2 |
|
daily_bw_url |
|
daily_served_url |
|
where
- grafanaHost
is the FQDN of the Grafana server (again, usually the same as the Traffic Stats server),
- cdn-stats-dashboard
is the name of the Dashboard providing CDN-level statistics,
- deliveryservice-stats-dashboard
is the name of the Dashboard providing Delivery Service-level statistics, and
- daily-summary-dashboard
is the name of the Dashboard providing a daily summary of general statistics that would be of interest to administrators using Traffic Portal
InfluxDB Tools
Under the Traffic Stats source directory there is a directory called influxdb_tools
. These tools are meant to be used as one-off scripts to help a user quickly get new databases and continuous queries setup in InfluxDB. They are specific for Traffic Stats and are not meant to be generic to InfluxDB. Below is an brief description of each script along with how to use it.
create/create_ts_databases.go
This program creates all databases, retention policies, and continuous queries required by Traffic Stats.
Pre-Requisites
Go 1.7 or later
Configured
$GOPATH
environment variable
Options and Arguments
- --help
(Optional) Print usage information and exit (with a failure exit code for some reason)
- --replication N
(Optional) The number of nodes in the cluster (default: 3)
- --url URL
The InfluxDB server’s root URL - including port number, if required (default:
http://localhost:8086
)
- --user username
The name of the user to use when connecting to InfluxDB
sync/sync_ts_databases.go
This program is used to sync one InfluxDB environment to another. Only data from continuous queries is synced as it is down-sampled data and much smaller in size than syncing raw data. Possible use cases are syncing from production to development or syncing a new cluster once brought online.
Pre-Requisites
Go 1.7 or later
Configured
$GOPATH
environment variable
Options and Arguments
- --database database_name
(Optional) Specify the name of a specific database to sync (default: all databases)
- --days N
The number of days in the past to sync.
0
means ‘all’
- --help
(Optional) Print usage information and exit
- --source-password password
The password of the user named by
--source-user
- --source-url URL
(Optional) The URL of the InfluxDB instance _from_ which data will be copied (default:
http://localhost:8086
)
- --source-user username
The name of the user as whom the utility will connect to the source InfluxDB instance
- --target-password password
The password of the user named by
--target-user
- --target-url URL
(Optional) The URL of the InfluxDB instance _to_ which data will be copied (default:
http://localhost:8086
)
- --target-user username
The name of the user as whom the utility will connect to the target InfluxDB instance