Configure Logs for Kubernetes

Introduction

Log Management is a configurable feature that can be enabled or disabled at the client level through the portal.

Prerequisites

The Logs Management option must be enabled to use logs management. This can be done either during client creation or after the client has been created:
- Enable Log Management while creating a new client
- Enable Log Management for an existing client
While installing the Kubernetes 2.0 Agent, enable the Enable Logs option. For more details, refer to the Installation document.

Agent Manifest File Configuration

Default Behavior

The agent, by default, collects all pod logs and syslogs from all nodes in the respective Kubernetes cluster. These logs are exported to OpsRamp if:
- The enabled field under the logs section in the downloaded/copied manifest file is set to true.
- The Enable Log Management option is set to “Yes” on the client details page.
Logs can be viewed in the Logs UI portal, located at: Infrastructure > Logs.

Example logs section:

logs:
  enabled: true

User Configuration of Logs

Default ConfigMap

The default ConfigMap provides a baseline configuration for logs collection, allowing users to customize it later as needed.

To list the default Kubernetes ConfigMap, use the following command:

kubectl get configmap opsramp-logs-user-config -n <agent-installed-namespace>

Default ConfigMap Example:

apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config-tmp"
  labels:
    app: "opsramp-logs-user-config"
  namespace: opsramp-agent
data:
  logsConfig.yaml: |
    pods:
      enable: true
      additional_labels:   
      namespaces:
        include: 
        is_include_regex: false
        exclude: 
        is_exclude_regex: false
      log_level: "Unspecified"
      masking:
        - attribute_type: ""
          attribute_key: ""
          text: ""
          placeholder: ""
    node:
      - logSource: syslog
        enable: true
        configs:
          - mount_path: ["/var/log/syslog","/var/log/messages"]         
            multiline:
              line_start_pattern: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)'
            parser_type: "regex"
            parser_expression: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)\s*(?P<syslog_tag>[^:]*):\s*(?P<message>.*)$'
            timestamp_layout_type: "strptime"
            timestamp_layout: "%b %e %H:%M:%S"

Disable Specific Logs: To disable either pod logs or syslogs, set enable: false under the respective pods or node sections in the opsramp-logs-user-config ConfigMap. The agent will automatically apply the changes and restart the OpenTelemetry collector.

Pods Logs Configuration

By default, the agent collects all pod logs. However, users can customize the log collection process using the opsramp-logs-user-config ConfigMap, which is automatically available after the agent is installed.

Default ConfigMap

apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config-tmp"
  labels:
    app: "opsramp-logs-user-config"
  namespace: opsramp-agent
data:
  logsConfig.yaml: |
    pods:
      enable: true
      additional_labels:   
      namespaces:
        include: 
        is_include_regex: false
        exclude: 
        is_exclude_regex: false
      log_level: "Unspecified"
      masking:
        - attribute_type: ""
          attribute_key: ""
          text: ""
          placeholder: ""

Customization Options

If users prefer not to collect pod logs by default, they can disable this feature by setting enable to false in the configuration. This can be done by running the following command:

kubectl edit configmap opsramp-logs-user-config -n <agent-installed-namespace>

Adding Additional Labels

The additional_labels field in the ConfigMap allows users to specify key-value pairs that will be added as attributes to each log record.

Steps to Configure:
Run the following command to edit the ConfigMap:

kubectl edit configmap opsramp-logs-user-config -n <agent-installed-namespace>

Example ConfigMap with Additional Labels:

apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config"
  labels:
    app: "opsramp-logs-user-config"
  namespace: <agent-installed-namespace>
data:
  logsConfig.yaml: |
    pods:
      enable: true
      additional_labels:  
        app: nginx
        env: staging

Explanation:
Here, app: nginx and env: staging are added as labels to all log records.

Namespace Filtering

By default, the agent collects pod logs from all namespaces. To restrict log collection to specific namespaces, users need to define inclusion or exclusion rules within the opsramp-logs-user-config ConfigMap.

Steps to Configure:

Run the following command to edit the ConfigMap:

kubectl edit configmap opsramp-logs-user-config -n <agent-installed-namespace>

Update the namespaces section in the logsConfig.yaml to specify the inclusion and exclusion rules. Example ConfigMap for Namespace Filtering
```
apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config"
  labels:
    app: "opsramp-logs-user-config"
  namespace: <agent-installed-namespace>
data:
  logsConfig.yaml: |  
    pods: 
      enable: true
      additional_labels:
      namespaces:
          include: 
            - ^monitoring
          is_include_regex: true
          exclude: 
            - monitoring-test
          is_exclude_regex: false
```
Explanation of Fields:
- The include and exclude fields can accept either a string or a regex pattern as a namespace.
- If a user specifies a regex pattern for a namespace in the include/exclude field, the corresponding flags is_include_regex or is_exclude_regex must be set to true.
- If a user specifies a namespace that is not a regex pattern in the include/exclude field, the flags is_include_regex and is_exclude_regex must be set to false.
- For example, the agent collects logs of pods in all namespaces that start with monitoring but excludes logs from pods in the monitoring-test namespace.

Use Case 1:

Inclusion	Exclusion	Result
Not set	Not set	Logs of all namespaces will be collected
Some filter	Not set	Logs of namespaces that match the inclusion filter
Not set	Some filter	Logs of all namespaces except those that match the exclusion filter
Some filter	Some filter	Logs of namespaces that match the inclusion filter minus logs that match the exclusion filter

Use Case 2:

Use Case	Inclusion	Exclusion
Collects logs of a single namespace " agent."	include: agent is_include_regex: false	-
Collect logs of all namespaces which starting with test only	include: ^test is_include_regex: true	-
Collects logs of all namespaces except the "kube-system" namespace	-	exclude: kube-system is_exclude_regex: false
Collects logs of all namespaces except namespace ending with the name " system."	-	exclude: system$ is_exclude_regex: true
Collects logs of all namespaces starting with test or kube.	include: ^test\|^kube is_include_regex: true	-
Collects logs of all namespaces which start with test or kube but exclude testing or kube-system namespaces	include: ^test\|^kube is_include_regex: true	exclude: testing, kube-system is_exclude_regex: false
Collects logs of namespaces which end with "agent" but excludes any namespace starting with opsramp	include: agent$ is_include_regex: true	exclude: ^opsramp is_exclude_regex: true

Log-Level Filtering

Log severity levels are ordered as follows: Fatal > Error > Warn > Info > Debug > Trace > Unspecified/Unknown.
By default, the agent evaluates log severity for each entry and sends logs to the OpsRamp platform based on this hierarchy.
Users can filter or drop logs based on their severity level by modifying the opsramp-logs-user-config ConfigMap.

Steps to Configure:
Run the following command to edit the ConfigMap:

kubectl edit configmap opsramp-logs-user-config -n <agent-installed-namespace>

Example 1: Filter logs with severity level Warn or higher (includes Warn, Fatal, and Error) and drop logs below Warn (Info, Debug, Trace, Unknown).

apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config-tmp"
  labels:
    app: "opsramp-logs-user-config"
  namespace: opsramp-agent
data:
  logsConfig.yaml: |
    pods:
      enable: true
      additional_labels:   
      namespaces:
        include: 
        is_include_regex: false
        exclude: 
        is_exclude_regex: false
      log_level: "Warn"

Example 2: Filter logs with severity level Info or higher (includes Info, Warn, Error, and Fatal) and drop logs below Info (Debug, Trace, Unknown).

apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config-tmp"
  labels:
    app: "opsramp-logs-user-config"
  namespace: opsramp-agent
data:
  logsConfig.yaml: |
    pods:
      enable: true
      additional_labels:   
      namespaces:
        include: 
        is_include_regex: false
        exclude: 
        is_exclude_regex: false
      log_level: "Info"

Note

The log_level field only accepts string values and is case-insensitive.
Allowed values: Fatal, Error, Warn, Info, Debug, Trace.

Masking

Masking involves obscuring sensitive or personally identifiable information (PII) in log entries to ensure compliance with data protection regulations.

Steps to Configure:
Run the following command to edit the ConfigMap:

kubectl edit configmap opsramp-logs-user-config -n <agent-installed-namespace>

Example Configuration:

apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config-tmp"
  labels:
    app: "opsramp-logs-user-config"
  namespace: opsramp-agent
data:
  logsConfig.yaml: |
    pods:
      enable: true
      additional_labels:   
      namespaces:
        include: 
        is_include_regex: false
        exclude: 
        is_exclude_regex: false
      log_level: "Unspecified"
      masking:
        - attribute_type: ""
          attribute_key: "k8s.namespace.name"
          text: "^ops"
          placeholder: "***"

Example Output:

To mask sensitive data (e.g., namespaces starting with “ops”):
If the namespace is opsramp-agent, it will appear as ***ramp-agent after masking.

Masking Field Descriptions:

	Description
attribute_type	Specifies the type of attribute being masked. Valid values are: resource and record
attribute_key	Specifies the name of the attribute key whose value requires masking.
text	Regular Expression for the text that needs to be masked.
placeholder	String which will be used for replacing the text specified.

Node Logs Configuration

By default, the agent collects node logs, such as syslogs, from all Kubernetes nodes using the opsramp-logs-user-config ConfigMap. This ConfigMap is available immediately after the agent installation.

Default ConfigMap

The following is the default configuration for syslog collection:

apiVersion: v1
kind: ConfigMap
metadata:
  name: "opsramp-logs-user-config-tmp"
  labels:
    app: "opsramp-logs-user-config"
  namespace: opsramp-agent
data:
  logsConfig.yaml: |
    node:
      - logSource: syslog
        enable: true
        configs:
          - mount_path: ["/var/log/syslog","/var/log/messages"]         
            multiline:
              line_start_pattern: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)'
            parser_type: "regex"
            parser_expression: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)\s*(?P<syslog_tag>[^:]*):\s*(?P<message>.*)$'
            timestamp_layout_type: "strptime"
            timestamp_layout: "%b %e %H:%M:%S"

Customization Options

If you do not want to collect syslogs by default, set enable: false in the configuration by running the following command:

kubectl edit configmap opsramp-logs-user-config -n <agent-installed-namespace>

Parser Expression Based on OS Version

For nodes running Ubuntu version 22.04:

parser_expression: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)\s*(?P<syslog_tag>[^:]*):\s*(?P<message>.*)$'

For nodes running Ubuntu version 24.0x, update to:

parser_expression: '^(?P<timestamp>\d{4}\W*\d{2}\W*\d{2}\w\d{2}:\d{2}:\d{2}.\d*\+\d{2}:\d{2})\s(?P<host>\w*\W\w*)\s(?P<syslog_tag>\w*\W\d*\W):(?P<message>.*)'

Configurable Fields

Field	Default	Description
logSource	syslog	The designated source name for collecting logs.
enable	true	This option is set to true by default. If the user wishes to disable it, they can change it to false.
configs	required	An array of configurations that specify the parsing logic.
mount_path	["/var/log/syslog","/var/log/messages"]	Indicates the file paths from which logs should be collected. Modify if a custom path is needed.
multiline	multiline: line_start_pattern: '^\s(?P\w\s\d\s\d:\d:\d)\s(?P[^\s])'	Syslogs default line_start_pattern. Directs the file_input operator to separate log entries based on a pattern other than newlines. Must include exactly one of line_start_pattern or line_end_pattern.
parser_type	regex	Can be either regex or json.
parser_expression	'^\s(?P\w\s\d\s\d:\d:\d)\s(?P[^\s])\s(?P[^:]):\s(?P.)$'	Syslogs default parser expression. Specifies the expression used to parse log records accordingly.
timestamp_layout_type	strptime	Valid values include strptime, gotime, and epoch.
timestamp_layout	"%b %e %H:%M:%S"	Defines the precise layout of the timestamp to be parsed.

Here are a few key points to keep in mind:

All fields in parser_expression will be parsed to log attributes only by default.
Users can customize all fields according to their preferences. Furthermore, they have the flexibility to modify the default configuration, as long as these settings are not the original defaults.
If a user has any custom configurations for syslog, in addition to the default settings, they can include these configurations in the ‘configs’ field as an array.

Example ConfigMap

node:
  - logSource: syslog
    enable: true
    configs:
      - mount_path: ["/var/log/syslog","/var/log/messages"]         
        multiline:
          line_start_pattern: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)'
        parser_type: "regex"
        parser_expression: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)\s*(?P<syslog_tag>[^:]*):\s*(?P<message>.*)$'
        timestamp_layout_type: "strptime"
        timestamp_layout: "%b %e %H:%M:%S"
      - mount_path: ["/opt/log/syslog/0.log"]         
        multiline:
          line_end_pattern: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)'
        parser_type: "json"
        parser_expression: '^\s*(?P<timestamp>\w*\s*\d*\s*\d*:\d*:\d*)\s*(?P<host>[^\s]*)\s*(?P<syslog_tag>[^:]*):\s*(?P<message>.*)$'
        timestamp_layout_type: "strptime"
        timestamp_layout: "%b %e %H:%M:%S"

Custom Path Configuration

By default, the agent has access to /var/log. If your log file is outside this directory, specify an alternative mount path using Helm:

helm upgrade <release-name> <agent-chart> --version 0.1.0 --reuse-values --set agent.worker.mountHostPath="{/var/log,/var/mnt}"

Note

The /var/log path is crucial for collecting pod logs. If you have custom paths, list them after /var/log, separated by commas (e.g., /var/mnt as a custom path). Also, make sure the --version parameter matches the currently installed Helm version.

View Logs in OpsRamp Portal

Users can view logs in the OpsRamp portal from Infrastructure > Logs.

Default Logs Screen: Access your logs through the Logs UI.
Logs Filter: Click on +FILTER to select resource types and operators (e.g., Equals, Regex Not Contains).
Live Tail: Click on LIVE TAIL to view real-time logs.
Log Details: Click on any log entry for detailed information.

Once you edit the K8s ConfigMap and apply it to your cluster, the OpsRamp Agent will pick up the latest configuration changes within a few minutes. This version organizes the content into clear sections and uses straightforward language for better understanding.