deliveryservice_stats
GET
Retrieves time-aggregated statistics on a specific Delivery Service.
- Auth. Required:
Yes
- Roles Required:
None[1]
- Permissions Required:
STAT:READ, DELIVERY-SERVICE:READ
- Response Type:
Object
Request Structure
Name |
Required |
Description |
---|---|---|
deliveryService |
yes[2] |
Either the xml_id of a Delivery Service for which statistics will be aggregated or the integral, unique identifier of said Delivery Service |
deliveryServiceName |
yes[2] |
The xml_id of the Delivery Service for which statistics will be aggregated |
endDate |
yes |
The date and time until which statistics shall be aggregated in RFC 3339 format (with or without sub-second precision), the number of nanoseconds since the Unix
Epoch, or in the same, proprietary format as the |
exclude |
no |
Either “series” to omit the data series from the result, or “summary” to omit the summary data from the result - directly corresponds to fields in the Response Structure |
interval |
no |
Specifies the interval within which data will be “bucketed”; e.g. when requesting data from 2019-07-25T00:00:00Z to 2019-07-25T23:59:59Z with an interval of “1m”,
the resulting data series (assuming it is not excluded) should contain
\(24\frac{\mathrm{hours}}{\mathrm{day}}\times60\frac{\mathrm{minutes}}{\mathrm{hour}}\times1\mathrm{day}\times1\frac{\mathrm{minute}}{\mathrm{data point}}=1440\mathrm{data\;points}\)
The allowed values for this parameter are valid InfluxQL duration literal strings matching |
limit |
no |
A natural number indicating the maximum amount of data points should be returned in the |
metricType |
yes |
The metric type being reported - one of:
|
offset |
no |
A natural number of data points to drop from the beginning of the returned data set |
orderby |
no |
Though one struggles to imagine why, this can be used to specify “time” to sort data points by their “time” (which is the default behavior) |
startDate |
yes |
The date and time from which statistics shall be aggregated in RFC 3339 format (with or without sub-second precision), the number of nanoseconds since the Unix
Epoch, or in the same, proprietary format as the |
GET /api/4.0/deliveryservice_stats?deliveryServiceName=demo1&startDate=2019-07-22T17:55:00Z&endDate=2019-07-22T17:56:00.000Z&metricType=tps_total HTTP/1.1
User-Agent: python-requests/2.20.1
Accept-Encoding: gzip, deflate
Accept: application/json;timestamp=unix, application/json;timestamp=rfc;q=0.9, application/json;q=0.8, */*;q=0.7
Connection: keep-alive
Cookie: mojolicious=...
Content Format
It’s important to note in Request Example the use of a complex “Accept” header. This endpoint accepts two special media types in the “Accept” header that instruct it on how to format the timestamps associated with the returned data. Specifically, Traffic Ops will recognize the special, optional, non-standard parameter of application/json: timestamp
. The values of this parameter are restricted to one of
- rfc
Returned timestamps will be formatted according to RFC 3339 (no sub-second precision).
- unix
Returned timestamps will be formatted as the number of nanoseconds since the Unix Epoch (midnight on January 1st 1970 UTC).
Implementation Detail
The endpoint passes back nanoseconds, specifically, because that is the form used both by InfluxDB, which is used to store the data being served, and Go’s standard library. Clients may need to convert the value to match their own standard libraries - e.g. the
Date()
class in Javascript expects milliseconds.
The default behavior - when only e.g. application/json or */* is given - is to use RFC 3339 formatting. It will, however, respect quality parameters. It is suggested that clients request timestamps they can handle specifically, rather than relying on this default behavior, as it is subject to change and is in fact expected to invert in the next major release as string-based time formats become deprecated.
See also
For more information on the “Accept” HTTP header, consult its dedicated page on MDN.
Response Structure
- series:
An object containing the actual data series and information necessary for working with it.
- columns:
This is an array of names of the columns of the data contained in the “values” array - should always be
["time", "sum_count"]
- count:
The number of data points contained in the “values” array
- name:
The name of the data set. Should always match
metric.ds.1min
wheremetric
is the requestedmetricType
- values:
The actual array of data points. Each represents a length of time specified by the
interval
query parameter- time:
The time at which the measurement was taken. This corresponds to the beginning of the interval. This time comes in the format of either an RFC 3339-formatted string, or a number containing the number of nanoseconds since the Unix Epoch depending on the “Accept” header sent by the client, according to the rules outlined in Content Format.
- value:
The value of the requested
metricType
at the time given bytime
. This will always be a floating point number, unless no data is available for the data interval, in which case it will benull
- source:
A legacy field meant only for plugins that override this endpoint to name themselves. Should always be “TrafficStats”.
Deprecated since version 1.4: As this has no known purpose, developers are advised it will be removed in the future.
- summary:
An object containing summary statistics describing the data series
- average:
The arithmetic mean of the data’s values
- count:
The number of measurements taken within the requested timespan. This is, in general, not the same as the
count
field of theseries
object, as it reflects the number of underlying, un-“bucketed” data points, and is therefore dependent on the implementation of Traffic Stats.- fifthPercentile:
Data points with values less than or equal to this number constitute the “bottom” 5% of the data set
- max:
The maximum value that can be found in the requested data set
- min:
The minimum value that can be found in the requested data set
- ninetyEighthPercentile:
Data points with values greater than or equal to this number constitute the “top” 2% of the data set
- ninetyFifthPercentile:
Data points with values greater than or equal to this number constitute the “top” 5% of the data set
- totalBytes:
When the
metricType
requested iskbps
, this will contain the total number of bytes transferred by the Delivery Service within the requested time window. Note that fractional amounts are possible, as the data transfer rate will almost certainly not be cleanly divided by the requested time range.- totalTransactions:
When the
metricType
requested is notkbps
, this will contain the total number of transactions completed by the Delivery Service within the requested time window. Note that fractional amounts are possible, as the transaction rate will almost certainly not be cleanly divided by the requested time range.
- version:
A legacy field that seems to have been meant to indicate the API version used. Will always be “1.2”
Deprecated since version 1.4: As this has no known purpose, developers are advised it will be removed in the future.
HTTP/1.1 200 OK
Access-Control-Allow-Credentials: true
Access-Control-Allow-Headers: Origin, X-Requested-With, Content-Type, Accept, Set-Cookie, Cookie
Access-Control-Allow-Methods: POST,GET,OPTIONS,PUT,DELETE
Access-Control-Allow-Origin: *
Content-Encoding: gzip
Content-Type: application/json
Set-Cookie: mojolicious=...; Path=/; Expires=Mon, 18 Nov 2019 17:40:54 GMT; Max-Age=3600; HttpOnly
Whole-Content-Sha512: zXJGjcYuu6HxWINVp8HA1gL31J3ukry5wCsTDNxtP/waC6rSD8h10KJ9jEAtRzJ9owOSVPvKaA/2bRu/QeuCpQ==
X-Server-Name: traffic_ops_golang/
Date: Mon, 22 Jul 2019 17:57:14 GMT
Transfer-Encoding: chunked
{ "response": {
"series": {
"columns": [
"time",
"sum_count"
],
"count": 2,
"name": "tps_total.ds.1min",
"tags": {
"cachegroup": "total"
},
"values": [
[
1563818100000000000,
0
],
[
1563818160000000000,
0
]
]
},
"source": "TrafficStats",
"summary": {
"average": 0,
"count": 2,
"fifthPercentile": 0,
"max": 0,
"min": 0,
"ninetyEighthPercentile": 0,
"ninetyFifthPercentile": 0,
"totalBytes": null,
"totalTransactions": 0
},
"version": "1.2"
}}