Readiness probe
This section describes how to configure readiness probes for your Fluentd and syslog-ng pods. If you don’t configure custom readiness probes, Logging operator uses the default probes.
Prerequisites
- Configuring readiness probes requires Logging operator 3.14.0 or newer installed on the cluster.
-
syslog-ng is supported only in Logging operator 4.0 or newer.
Overview of default readiness probes
By default, Logging operator performs the following readiness checks:
- Number of buffer files is too high (higher than 5000)
- Fluentd buffers are over 90% full
- syslog-ng buffers are over 90% full
The parameters of the readiness probes and pod failure is set by using the usual Kubernetes probe configuration parameters. Instead of the Kubernetes defaults, the Logging operator uses the following values for these parameters:
Currently, you cannot modify the default readiness probes, because they are generated from the source files. For the detailed list of readiness probes, see the Default readiness probes. However, you can customize their values in the Logging custom resource, separately for the Fluentd and syslog-ng log forwarder. For example:
Fluentd readiness probe settings
SyslogNG readiness probe settings
Default readiness probes
The Logging operator applies the following readiness probe by default:
Add custom readiness probes
You can add your own custom readiness probes to the spec.ReadinessProbe section of the logging custom resource. For details on the format of readiness probes, see the official Kubernetes documentation.
CAUTION:
If you set any custom readiness probes, they completely override the default probes.