affinity #203
affinity #203
Conversation
docs/use/affinity.md
Outdated
| @@ -0,0 +1,170 @@ | |||
| # Leveraging Affinity in Percona Everest | |||
|
|
|||
| !!! warning | |||
There was a problem hiding this comment.
Maybe it should just be an info or note instead of a warning
There was a problem hiding this comment.
@recharte Let's say it is Important.
docs/use/affinity.md
Outdated
|
|
||
| |**Attributes** |**Description** |**Note** | | ||
| |----------------|-----------------|--------------| | ||
| |**Components**|<ul><li>Config Server</li><li>DB Node</li><li>Router</br></br></li><li>DB Node</br><br></li><li>DB Node</li><li>Proxy</li></ul>|Applicable for **MongoDB sharded cluster**.<br/></br></br></br></br>Applicable for **MongoDB non sharded cluster**.</br></br>Applicable for **MySQL** and **PostgreSQL** databases.| |
There was a problem hiding this comment.
I don't feel strongly opinionated about this but I think we could make the table simpler to understand if we go with something like:
- DB node
- Proxy/Router/PGBouncer - Applicable for MySQL, MongoDB sharded clusters and PostgreSQL databases.
- Config Server - Applicable for MongoDB sharded clusters.
docs/use/affinity.md
Outdated
| ```sh | ||
| affinity: | ||
| nodeAffinity: | ||
| requiredDuringSchedulingIgnoredDuringExecution: | ||
| nodeSelectorTerms: | ||
| - matchExpressions: | ||
| - key: topology.kubernetes.io/region | ||
| operator: In | ||
| values: | ||
| - "us-west2" | ||
| ``` |
There was a problem hiding this comment.
Shouldn't we show a screenshot configuring this instead?
There was a problem hiding this comment.
I didnt have the build so instead added this...
docs/use/affinity.md
Outdated
| Let's consider a use case in which you want to ensure that no HAProxy pods should be scheduled to run on the same Kubernetes node. | ||
|
|
||
|
|
||
| affinity: | ||
| podAntiAffinity: | ||
| requiredDuringSchedulingIgnoredDuringExecution: | ||
| - labelSelector: | ||
| matchExpressions: | ||
| - key: app | ||
| operator: NotIn | ||
| values: | ||
| - haproxy | ||
| topologyKey: "kubernetes.io/hostname" | ||
|
|
||
| ??? info "What happens under the hood" | ||
| - The pod will not be scheduled on nodes that contain pods labeled with `app=haproxy`. | ||
|
|
||
| - If no nodes match, the pod will not be scheduled until a suitable node becomes available. |
There was a problem hiding this comment.
Why don't we use the default rule that we set as an example?
Simply setting topologyKey: "kubernetes.io/hostname" ensures that the scheduler will try to spread the different DB pods across different k8s nodes, ensuring better fault tolerance in case of node failure.
There was a problem hiding this comment.
I also think it's important to have the default rule explained somewhere so we could do both here 🙂
There was a problem hiding this comment.
We can explain it here itself.
There was a problem hiding this comment.
Definitely. It will be more useful.
Co-authored-by: Diogo Recharte <[email protected]>
rasika-chivate
temporarily deployed
to
affinity - Everest Release Branch PR #203
Render
Destroyed
rasika-chivate
temporarily deployed
to
affinity - Everest Release Branch PR #203
Render
Destroyed
Co-authored-by: Diogo Recharte <[email protected]>
rasika-chivate
temporarily deployed
to
affinity - Everest Release Branch PR #203
Render
Destroyed
Co-authored-by: Diogo Recharte <[email protected]>
rasika-chivate
temporarily deployed
to
affinity - Everest Release Branch PR #203
Render
Destroyed
Co-authored-by: Diogo Recharte <[email protected]>
Co-authored-by: Diogo Recharte <[email protected]>
No description provided.