Cloud Native Logging Add-on

Overview

log-agent is an add-on based on Fluent Bit and OpenTelemetry for cloud native logging. It supports CRD-based log collection policies, as well as collects and forwards standard output logs, container file logs, node logs, and Kubernetes events of containers in a cluster. After the add-on is installed, standard output logs and Kubernetes events are collected by default. For details about how to use log-agent to collect logs, see Collecting Data Plane Logs.

Constraints

Permissions

The fluent-bit component of the log-agent add-on reads and collects the standard output logs on each node, file logs in pods, and node logs based on the collection configuration.

The following permissions are required for running the fluent-bit component:

• CAP_DAC_OVERRIDE: ignores the discretionary access control (DAC) restrictions on files.
• CAP_FOWNER: ignores the restrictions that the file owner ID must match the process user ID.
• DAC_READ_SEARCH: ignores the DAC restrictions on file reading and catalog research.
• SYS_PTRACE: allows all processes to be traced.

Assigning Authorization for log-agent in Your On-Premises Cluster

The log-agent add-on needs to be authenticated before accessing LTS and AOM. This add-on uses Workload Identity to allow workloads in your on-premises cluster to impersonate IAM users to access cloud services.

Workload Identity allows you to configure the public key of your cluster for the IAM IdP and add a mapping rule to map a ServiceAccount to an IAM service account. During workload deployment, the token of the ServiceAccount is mounted to the workload. This token is used to access cloud services. This way, the AK/SK of the IAM service account is not required, reducing security risks.

1. Obtain the JSON Web Key Set (JWKS) of the on-premises cluster, which is used to verify the ServiceAccount token issued by ClusterIssuer.
   1. Use kubectl to access the on-premises cluster.
   2. Run the following command to obtain the public key:

      kubectl get --raw /openid/v1/jwks

      A json string is returned, containing the signature public key of the cluster for accessing the IdP.

      {
          "keys": [
              {
                  "kty": "RSA",
                  "e": "AQAB",
                  "use": "sig",
                  "kid": "Ew29q....",
                  "alg": "RS256",
                  "n": "peJdm...."
              }
          ]
      }

2. Configure an IdP entity for your on-premises cluster on IAM.
   1. Log in to the IAM console, query the ID of the project that the on-premises cluster belongs to, create an identity provider, and select OpenID Connect for Protocol. Enter the IdP name for log-agent. For details, see Table 1. For details about how to configure permissions for a user group, see User Group Policy Content.

      Table 1 log-agent IdP settings

      Add-on Name	IdP Name	Client ID	Namespace	ServiceAccount Name	Minimum Permissions on User Groups
      log-agent	ucs-cluster-identity-{Project ID}	ucs-cluster-identity	monitoring	log-agent-serviceaccount	aom:alarm:* lts:*:*

      Figure 1 Modifying identity provider information
      
   2. Click OK and modify the IdP information as described in Table 2. Click Create Rule to create an identity conversion rule.
      Figure 2 Modifying identity provider information
      

      Table 2 IdP parameters

      Parameter	Description
      Access Type	Select Programmatic access.
      Configuration Information	Identity Provider URL: Enter https://kubernetes.default.svc.cluster.local. Client ID: Enter the client ID of log-agent. For details, see Table 1. Signing Key: Enter the JWKS of the on-premises cluster obtained in 1. If multiple clusters are involved, use commas (,) to separate the keys array content of each cluster.
      Identity Conversion Rules	An identity conversion rule maps a ServiceAccount in an on-premises cluster to an IAM user group. Attribute: sub Condition: any_one_of Value: Value format: system:serviceaccount:Namespace:ServiceAccountName. Change Namespace to the namespace for which the ServiceAccount is to be created, and change ServiceAccountName to the name of the ServiceAccount to be created. For example, if the value is system:serviceaccount:monitoring:log-agent-serviceaccount, a ServiceAccount named log-agent is created in the monitoring namespace and mapped to the corresponding user group. The IAM token obtained using this ServiceAccount has the permissions of the user group. NOTE: ServiceAccountName and user group permissions are mandatory for running add-ons in an on-premises cluster. For details, see Table 1.

      Figure 3 Creating an identity conversion rule
      
   3. Click OK.

Installing log-agent in an On-Premises Cluster

1. Log in to the UCS console and choose Fleets. Then, click the cluster name to access the cluster console. In the navigation pane, choose Add-ons. Locate Cloud Native Logging on the right and click Install.
2. On the Install Add-on page, configure the specifications.

   Table 3 Add-on specifications

   Parameter	Description
   Add-on Specifications	The add-on specifications can be of the Low, High, or custom-resources type.
   Pods	Number of pods that will be created to match the selected add-on specifications. If you select custom-resources, you can adjust the number of pods as required.
   Containers	The log-agent add-on contains the following containers, whose specifications can be adjusted as required: fluent-bit: indicates the log collector, which is installed on each node as a DaemonSet. cop-logs: generates and updates configuration files on the collection side. log-operator: parses and updates log collection rules. otel-collector: forwards logs collected by fluent-bit to LTS in a centralized manner.

3. Configure the parameters in Parameters.

   Interconnection with AOM: If this option is enabled, Kubernetes events will be collected and reported to AOM. You can configure alarm rules on AOM.

4. Configure the network for reporting add-on instance logs.
   • Public network: This option features flexibility, cost-effectiveness, and easy access. It is only available for clusters that can access the public network.
   • Direct Connect or VPN: After you connect an on-premises data center to a VPC over Direct Connect or VPN, you can use a VPC endpoint to access CIA over the private network. This option features high speed, low latency, and high security. For details, see Using Direct Connect or VPN to Report Logs of On-Premises Clusters.

5. Click Install.

log-agent Components

Change History

Reporting Custom Events to AOM

The log-agent add-on reports all warning events and some normal events to AOM. You can also set the events to be reported as required.

1. Run the following command on the cluster to modify the event collection settings:

   kubectl edit logconfig -n kube-system default-event-aom

2. Modify the event collection settings as required.

   apiVersion: logging.openvessel.io/v1
   kind: LogConfig
   metadata:
     annotations:
       helm.sh/resource-policy: keep
     name: default-event-aom
     namespace: kube-system
   spec:
     inputDetail:    # Settings on UCS from which events are collected
       type: event    # Type of logs to be collected. Do not change the value.
       event:
         normalEvents:    # Used to configure normal events
           enable: true    # Whether to enable normal event collection
           includeNames:    # Names of events to be collected. If this parameter is not specified, all events will be collected.
           - NotTriggerScaleUp
           excludeNames:    # Names of events that are not collected. If this parameter is not specified, all events will be collected.
           - NotTriggerScaleUp
         warningEvents:    # Used to configure warning events
           enable: true    # Whether to enable warning event collection
           includeNames:    # Names of events to be collected. If this parameter is not specified, all events will be collected.
           - NotTriggerScaleUp
           excludeNames:    # Names of events that are not collected. If this parameter is not specified, all events will be collected.
           - NotTriggerScaleUp
     outputDetail:
       type: AOM    # Type of the system that receives the events. Do not change the value.
       AOM:
         events:
         - name: DeleteNodeWithNoServer    # Event name. This parameter is mandatory.
           resourceType: Namespace    # Type of the resource that operations are performed on.
           severity: Major    # Event severity after an event is reported to AOM, which can be Critical, Major, Minor, or Info. The default value is Major.

log-agent Events

During log-agent installation and running, the log-operator component reports events. You can determine whether log-agent is installed and determine fault causes based on these events. For details, see Table 6.

log-agent Metrics

The log-operator, fluent-bit, and otel-collector components of the log-agent add-on have a series of metrics. You can use AOM or Prometheus to monitor these metrics to check the running of the log-agent add-on in a timely manner. For details, see Monitoring Custom Metrics Using AOM or Monitoring Custom Metrics Using Prometheus. The following lists the metrics:

• log-operator (only for Huawei Cloud clusters)

  Port: 8443

  Address: /metrics

  Protocol: HTTPS

  Table 7 Metrics

  Metric	Description	Type
  log_operator_aksk_latest_update_times	Last update time of the AK/SK	Gauge
  log_operator_aksk_update_total	Cumulative count of AK/SK update times	Counter
  log_operator_send_request_total	Cumulative count of requests that have been sent	Counter
  log_operator_webhook_listen_status	Webhook listening status	Gauge
  log_operator_http_request_duration_seconds	HTTP request latency	Histogram
  log_operator_http_request_total	Cumulative count of HTTP requests	Counter
  log_operator_webhook_request_total	Cumulative count of webhook requests	Counter

• fluent-bit

  Port: 2020

  Address: /api/v1/metrics/prometheus

  Protocol: HTTP

  Table 8 Metrics

  Metric	Description	Type
  fluentbit_filter_add_records_total	Cumulative count of records added by the Fluent Bit filter add-on	Counter
  fluentbit_filter_drop_records_total	Cumulative count of records dropped by the Fluent Bit filter add-on	Counter
  fluentbit_input_bytes_total	Number of input bytes	Counter
  fluentbit_input_files_closed_total	Cumulative count of files closed by the Fluent Bit input add-on	Counter
  fluentbit_input_files_opened_total	Cumulative count of files opened by the Fluent Bit input add-on	Counter
  fluentbit_input_files_rotated_total	Cumulative count of files rotated by the Fluent Bit input add-on	Counter
  fluentbit_input_records_total	Number of input records	Counter
  fluentbit_output_dropped_records_total	Number of dropped records	Counter
  fluentbit_output_errors_total	Number of output errors	Counter
  fluentbit_output_proc_bytes_total	Number of processed output bytes	Counter
  fluentbit_output_proc_records_total	Number of processed output records	Counter
  fluentbit_output_retried_records_total	Number of retried records	Counter
  fluentbit_output_retries_total	Number of output retries	Counter
  fluentbit_uptime	Number of seconds that Fluent Bit has been running	Counter
  fluentbit_build_info	Build version information	Gauge

• otel-collector

  Port: 8888

  Address: /metrics

  Protocol: HTTP

  Table 9 Metrics

  Metric	Description	Type
  otelcol_exporter_enqueue_failed_log_records	Number of log records failed to be added to the sending queue	Counter
  otelcol_exporter_enqueue_failed_metric_points	Number of metric points failed to be added to the sending queue	Counter
  otelcol_exporter_enqueue_failed_spans	Number of spans failed to be added to the sending queue	Counter
  otelcol_exporter_send_failed_log_records	Number of log records failed to be sent	Counter
  otelcol_exporter_sent_log_records	Number of log records that have been sent	Counter
  otelcol_process_cpu_seconds	Total CPU user and system time in seconds	Counter
  otelcol_process_memory_rss	Total physical memory (resident set size)	Gauge
  otelcol_process_runtime_heap_alloc_bytes	Bytes of allocated heap objects	Gauge
  otelcol_process_runtime_total_alloc_bytes	Cumulative bytes allocated for heap objects	Counter
  otelcol_process_runtime_total_sys_memory_bytes	Total bytes of memory obtained from the OS	Gauge
  otelcol_process_uptime	Uptime of the process in seconds	Counter
  otelcol_receiver_accepted_log_records	Number of log records received and processed by the OpenTelemetry receiver	Counter
  otelcol_receiver_refused_log_records	Number of log records rejected by the OpenTelemetry receiver	Counter