Skip to content

Common Tag Organization


In the data collected by DataKit, Tag is the key field of all data, which affects the filtering and grouping of data. Once the Tag data is wrong, it will lead to the wrong display of Web page data. In addition, Tag calibration will also affect the consumption statistics of time series data. Therefore, in the process of designing and changing Tag, we should think carefully and consider whether the corresponding changes will cause related problems. This document mainly lists the common tags in DataKit at present, so as to clarify the specific meaning of each Tag. Secondly, when adding new tags in the future, the following tags should be used and followed to avoid inconsistency.

The following will be listed from two dimensions: global Tag and specific data type Tag.

Global Tag

These tags are independent of the specific data type, and can be appended to any data type.

Tag Description
host Hostname, DaemonSet installation and host installation can all carry this tag, and in certain cases, users can rename the value of this tag.
project Project name, which is usually set by the user.
cluster Cluster name, usually set by the user in DaemonSet installation.
election_namespace The namespace of the election is not appended by default. See the document.
version Version number, all tag fields involving version information, should be represented by this tag.

Kubernetes/Common Tag of Container

These tags are usually added to the collected data, but when it comes to time series collection, some changeable tags (such as pod_name) will be ignored by default to save the timeline.

Tag Description
pod_name Pod name
deployment Deployment name in k8s
service Service name in k8s
namespace Namespace name in k8s
job Job name in k8s
image Full name of mirroring in k8s
image_name Abbreviation of mirror name in k8s
container_name K8s/Container name in the container
cronjob CronJob name in k8s
daemonset DaemonSet name in k8s
replica_set ReplicaSet name in k8s
node_name Node name in k8s
node_ip Node IP in k8s

Tag Categorization of Specific Data Types

Log

Tag Description
source The log source exists as a metric set name on the line protocol, not as a tag. The center stores it as a tag as the source field of the log.
service Referring to the service name of the log. If not filled in, its value is equivalent to the source field
status Referring to log level. If it is not filled in, the collector will set its value to unknown by default, and the common status list is here.

Object

Tag Description
class Referring to object classification. It exists as a metric set name on the row protocol, instead of a tag. But the center stores it as a tag as the class field of the object
name Referring to object name. The center combines hash (class + name) to uniquely identify objects in a workspace.

Metrics

There is no fixed tag except the global tags because of the various data sources.

APM

The tag of Tracing class data is unified here.

RUM

See RUM document.

Scheck

See the Scheck doc.

Profile

See the collector doc.

Network

See the collector doc.

Event

See the design doc.

Feedback

Is this page helpful? ×