Prometheus data features:
Prometheus stores the pulled data in its own DB, which is called TSDB (Time Series Database). Each indicator corresponds to a hash value as the time point changes, thus forming a time series, and multiple indicators constitute Multiple time series, users can extract indicator data and display charts according to their own needs;
In the prometheus expression language, there are four types:
instant vector: A set of time series containing a single sample of each time series, sharing the same timestamp.
Range vector: A set of time series containing data points for each time series over time.
Scalar: a simple numeric floating point value
String (String): a simple string value (currently not used)
Prometheus provides a functional expression language that allows users to select and aggregate time series data. The results of expression query can be drawn as a graph, or displayed as a table in the expression browser provided by prometheus, or can be called and used by an external system with HTTP API.
Expression is the writing specification of PromQl:
Prometheus uniquely defines a time series based on the metrics name and associated label set
- The name of the indicator represents the basic characteristic identification of a certain type of measurable attribute on the monitoring target
- Labels are multiple measurable dimensions subdivided from this basic feature
PromQl is the built-in data query language of prometheus Server
How to use PromQl?
Open the web page of prometheus, and enter the indicator name in the Expression box:
Data filtering:
Equivalent query:
Click excute to get the query results. If you want to further filter the data and narrow the scope, you can add brackets {} after the metric name and add label attributes in the brackets to filter: for example, node_cpu_seconds_total{cpu
="0"} is Select the data of the node_cpu_seconds_total indicator and the cpu is 0:
unequal query
When filtering with tags, you can also use inequality queries:
offset modifier
The offset modifier allows to change the time offset of a single instant vector and range vector in the query.
For example:
node_cpu_seconds_total{cpu="0", mode=~"s.*"}offset 5m
is to view the value of the indicator 5m ago:
Applications of regular expressions
Use =~"expression" to match regular expressions
such as: node_cpu_seconds_total{cpu="0", mode=~"s.*"} is to select all samples whose CPU is 0 and whose mode value starts with s value:
Queries for range vectors:
The use of the real-time vector displayed above, that is, the indicator value at a certain moment, will display the data within the last 1 hour by default in the Prometheus graph:
Add a square bracket [] after the expression in Expression, and the original instant vector becomes a range vector when the time range is defined in the brackets, such as 5m and 1h. At this time, the amount of data is not one, but multiple:
range The unit of the vector selector:
s:seconds
m:minutes h
:hours
d:days
w:weeks
y:years
, for example, the following expression returns the value of http_requests_total 5 minutes ago relative to the current query time:
Then the range vector cannot be drawn, because Prometheus has added a time range by default when the instant vector is drawn, and the range vector itself is a "bunch" of data with a time range, which cannot be displayed as a trend graph. At this time, it is generally It will be processed with the rate function to get the change rate of the node_cpu_seconds_total indicator within 5m minutes:
so that it can be drawn:
