Fluentd Logging¶
Overview¶
The Service Proxy for Kubernetes (SPK) Fluentd pod is an open source data collector that can be configured to receive logging data from the SPK Controller, Traffic Management Microkernel (TMM), and Distributed Session State Management (dSSM) pods. The logs collected by Fluentd can be routed to standard output, filesystem, and Elasticsearch. To store the logs on the filesystem, the Fluentd pod must be bound to a Kubernetes persistence volume.
This document guides you through configuring and deploying the Fluentd pod to collect and store logs from different SPK components.
Fluentd Service¶
Upon installing SPK Fluentd, a Kubernetes Service object is created to receive logging data from Fluent bit on TCP service port 54321 and forward the data to Fluentd on TCP target port 24224. Ensure the service port is available, and the cluster has CoreDNS enabled. In the following example setup, Fluentd is deployed in two projects to collect logs from their respective pods.
Example Fluentd Setup:
Example Fluentd Service:
Name: f5-toda-fluentd
Namespace: spk-utilities
Port: 54321/TCP
Name: f5-toda-fluentd
Namespace: spk-ingress
Port: 54321/TCP
Log file locations¶
The log data collected by Fluentd is stored in the following locations:
Container | Log file |
---|---|
f5-dssm-sentinel | /var/log/f5/f5-dssm-sentinel-0/sentinel.log |
f5-dssm-db | /var/log/f5/f5-dssm-db-0/dssm.log |
f5ingress | /var/log/f5/helm_release-f5ingress/pod_name/f5ingress.log |
f5-tmm | /var/log/f5/f5-tmm/pod_name/f5-fsm-tmm.log |
f5-tmm-routing | /var/log/f5/f5-tmm/pod_name/f5-tmm-routing.log |
Note: To modify the TMM logging level, review the bdt_cli section of the Debug Sidecar overview.
Requirements¶
Prior to installing Fluentd, ensure you have:
- An OpenShift cluster.
- An available persistence volume.
- Installed the SPK software.
- A Linux based workstation with Helm installed.
Procedures¶
Installation¶
Use the following steps to the install the f5-fluentd pod.
Change into local directory with the SPK files, and list the files in the tar directory:
In this example, the SPK files are in the spkinstall directory:
cd spkinstall
ls -1 tar
In this example, Fluentd Helm chart is named f5-toda-fluentd-3.0.25.tgz:
csrc-0.4.2.tgz cwc-2.0.16.tgz f5-cert-gen-0.5.2.tgz f5-cert-manager-0.5.12.tgz f5-dssm-3.0.37.tgz f5-toda-fluentd-3.0.25.tgz f5ingress-10.0.104.tgz rabbitmq-2.0.5.tgz spk-docker-images.tgz
Create two new projects for deploying Fluentd:
oc new-project <project>
Note: You could skip project creation step if you already have existing projects to deploy Fluentd, created as part of dSSM Database and SPK Controller.
In this example, two new projects named spk-utilities and spk-ingress are created:
oc new-project spk-utilities oc new-project spk-ingress
Create a Helm values file named fluentd-values.yaml, and set the
image.repository
and thepersistence.storageClass
parameters:image: repository: '<registry>' persistence: enabled: true storageClass: '<name>'
In this example, Helm pulls the f5-fluentd image from registry.com, and the pod will bind to the storageClass named managed-nfs-storage:
image: repository: 'local.registry.com' persistence: enabled: true storageClass: 'managed-nfs-storage'
In fluentd-values.yaml file set the
serviceAccount.create
parameter:Note: The serviceAccount will not be created by default.
serviceAccount: create: false name: default
Required: Add the following parameters to fluentd-values.yaml file to collect logging data from different SPK components:
Important: The Fluent bit and Fluentd configurations are mandatory for proper log file recovery in the event of pod or container restarts; without this configuration, the users will not be able to recover their log files.
# enable controller logging f5ingress_logs: enabled: true stdout: true # enable dSSM DB logging dssm_logs: enabled: true stdout: true # enable dSSM Sentinel logging dssm_sentinel_logs: enabled: true stdout: true # enable cert manager logging cm_logs: enabled: true stdout: true
Note: In this example, we are using the same values file for both spk-utilities and spk-ingress projects. It can be split into separate files (fluentd-spk-utilities-values.yaml and fluentd-spk-ingress-values.yaml) if required, to contain only the applicable components of the respective projects.
Install the f5-fluentd pod and reference the fluentd-values.yaml values file. Be certain to make note of the Fluentd hostnames displayed after the installation in order to update Fluent bit sidecar configuration in the last step:
In this example, the Fluentd Pod installs to the spk-utilities and spk-ingress projects.
helm install f5-fluentd tar/f5-toda-fluentd-3.0.25.tgz -f fluentd-values.yaml -n spk-utilities
Fluentd hostname: f5-toda-fluentd.spk-utilities.svc.cluster.local. Fluentd port: "54321"
helm install f5-fluentd tar/f5-toda-fluentd-3.0.25.tgz -f fluentd-values.yaml -n spk-ingress
Fluentd hostname: f5-toda-fluentd.spk-ingress.svc.cluster.local. Fluentd port: "54321"
The f5-fluentd pod should now be successfully installed:
In this example, the Fluentd Pod STATUS is Running:
oc get pods -n spk-ingress
NAME READY STATUS f5-toda-fluentd-8cf96967b-jxckr 1/1 Running
oc get pods -n spk-utilities
NAME READY STATUS f5-toda-fluentd-9cgg22gc-iuacd 1/1 Running
Fluentd should also be bound to the persistent volume:
In this example, the Fluentd Pod PVC displays STATUS as Bound:
oc get pvc -n spk-ingress
NAME STATUS VOLUME STORAGECLASS f5-toda-fluentd Bound pvc-7d36b530-b718-466c-9b6e-895e8f1079a2 managed-nfs-storage
oc get pvc -n spk-utilities
NAME STATUS VOLUME STORAGECLASS f5-toda-fluentd Bound pvc-4e77x331-u529-566w-6c1r-913f4g1116c1 managed-nfs-storage
Update Fluent bit sidecar of all pods to redirect logs to Fluentd pod deployed in their respective projects:
Note: In this example, the Fluentd hostnames are f5-toda-fluentd.spk-utilities.svc.cluster.local. and f5-toda-fluentd.spk-ingress.svc.cluster.local. The hostname value needs to be updated in dssm-values.yaml and ingress-values.yaml.
sentinel: fluentbit_sidecar: image: repository: 'local.registry.com' fluentd: host: 'f5-toda-fluentd.spk-utilities.svc.cluster.local.' db: fluentbit_sidecar: image: repository: 'local.registry.com' fluentd: host: 'f5-toda-fluentd.spk-utilities.svc.cluster.local.'
controller: fluentbit_sidecar: enabled: true fluentd: host: 'f5-toda-fluentd.spk-ingress.svc.cluster.local.' image: repository: 'local.registry.com' f5-toda-logging: enabled: true fluentd: host: 'f5-toda-fluentd.spk-ingress.svc.cluster.local.'
Viewing logs¶
After installing the Controller and dSSM pods, you can use the following steps to view the logs in the f5-fluentd
container:
Log in to the fluentd container:
oc exec -it deploy/f5-toda-fluentd -n <project> -- sh
In this example, the container is in the spk-utilities project:
oc exec -it deploy/f5-toda-fluentd -n spk-utilities -- sh
Change to the main logging directory, and list the subdirectories:
cd /var/log/f5; ls
In this example, logging directories are present for the f5-tmm, f5-dssm-db, and f5-dssm-sentinel pods:
f5-dssm-db-0 f5-dssm-db-1 f5-dssm-db-2 f5-dssm-sentinel-0 f5-dssm-sentinel-1 f5-dssm-sentinel-2 f5-tmm
Change into one of the subdirectories, for example f5-dssm-db-0:
cd f5-dssm-db-0
View the logs using the more command:
more -d dssm.log
Note: Logs collected by Fluentd are also available as part of QKView diagnostic tarball. To include log files from all Fluentd pods, remember to run the QKView utility with their respective namespaces.
In this example, QKView utility should be invoked as follows:
./qkview-wrapper-linux -f ./qkview-collector.sh -n spk-ingress,spk-utilities
(or)
./qkview-wrapper-darwin -f ./qkview-collector.sh -n spk-ingress,spk-utilities
In the iHealth website, the log files from different pods would be visible under their respective Fluentd pod’s filesystem subtree:
Next step¶
Continue to one of the following steps listed by installation precedence:
- Optional: Install the dSSM Database to store session-state information.
- Required: Install the SPK Controller and Service Proxy TMM pods.
Feedback¶
Provide feedback to improve this document by emailing spkdocs@f5.com.