Refactor probes based on performance feedback #264
Conversation
pega-talba
requested review from
dcasavant,
taz-pega-work and
a team
as code owners
There was a problem hiding this comment.
excellent new content summaries. I see that you updated the percentages to integers for maxSurge and maxUnavailable in several yamls...that detail for what it means isn't documented anywhere. I makes me wonder if that should be a future enhancement for the effected helm charts in charts/pega. Nothing in your updates.
Just wanted to note that we made the necessary changes as part of #261 when adding a default server.xml to the config/deploy directory. |
pega-talba
added
the
integ-all
|
Starting PR-264 validation on -> integ-all |
charts/pega/README.md
Outdated
| [Probes are used by Kubernetes](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/) to determine application health. Configure a probe for *liveness* to determine if a Pod has entered a broken state; configure it for *readiness* to determine if the application is available to be exposed; configure it for *startup* to determine if a pod is ready to be checked for liveness. You can configure probes independently for each tier. If not explicitly configured, default probes are used during the deployment. Set the following parameters as part of a `livenessProbe`, `readinessProbe`, or `startupProbe` configuration. | ||
|
|
||
| Notes: | ||
| * `startupProbe` is only supported as of Kubernetes 1.18. If running a version older than 1.18, `startupProbe` will be ignored and different default values will be used for `livenessProbe` and `readinessProbe`. |
There was a problem hiding this comment.
Kubernetes 1.18 and later supports startupProbe. If your deployment uses a Kubernetes version older than 1.18, it ignores your startupProbe settings and uses different default values for livenessProbe and readinessProbe.
There was a problem hiding this comment.
Done (with minor modifications)
charts/pega/README.md
Outdated
|
|
||
| Notes: | ||
| * `startupProbe` is only supported as of Kubernetes 1.18. If running a version older than 1.18, `startupProbe` will be ignored and different default values will be used for `livenessProbe` and `readinessProbe`. | ||
| * `timeoutSeconds` cannot be greater than `periodSeconds` in some GCP environments. See [this API library from Google](https://developers.google.com/resources/api-libraries/documentation/compute/v1/csharp/latest/classGoogle_1_1Apis_1_1Compute_1_1v1_1_1Data_1_1HttpHealthCheck.html#a027a3932f0681df5f198613701a83145). |
There was a problem hiding this comment.
For details, see [this API library from Google]
charts/pega/README.md
Outdated
| * `timeoutSeconds` cannot be greater than `periodSeconds` in some GCP environments. See [this API library from Google](https://developers.google.com/resources/api-libraries/documentation/compute/v1/csharp/latest/classGoogle_1_1Apis_1_1Compute_1_1v1_1_1Data_1_1HttpHealthCheck.html#a027a3932f0681df5f198613701a83145). | ||
| * Default values are listed below in order of liveness, readiness, and startup. | ||
|
|
||
| Parameter | Description | Default - 1.18+ | Default - pre 1.18 |
charts/pega/README.md
Outdated
| `failureThreshold` | The number consecutive failures for the pod to be terminated by Kubernetes. | `3` | ||
| ### Liveness, readiness, and startup probes | ||
|
|
||
| [Probes are used by Kubernetes](https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-startup-probes/) to determine application health. Configure a probe for *liveness* to determine if a Pod has entered a broken state; configure it for *readiness* to determine if the application is available to be exposed; configure it for *startup* to determine if a pod is ready to be checked for liveness. You can configure probes independently for each tier. If not explicitly configured, default probes are used during the deployment. Set the following parameters as part of a `livenessProbe`, `readinessProbe`, or `startupProbe` configuration. |
There was a problem hiding this comment.
Pega supports using liveness, readiness, and startup probes to determine application health in your deployments. For an overview of these probes, see Configure Liveness, Readiness and Startup Probes .
There was a problem hiding this comment.
Done (with minor modifications)
|
Starting PR-264 validation on -> integ-all |
pega-talba
added
integ-gke
|
Starting PR-264 validation on -> integ-gke |
|
Starting PR-264 validation on -> integ-gke |
No description provided.