Replace kamaji-etcd with aenix-io/etcd-operator #95
Conversation
kvaps
force-pushed
the
etcd-operator
branch
4 times, most recently
from
25bb31c
to
c5f1c85
Compare
kvaps
changed the title
kvaps
force-pushed
the
etcd-operator
branch
2 times, most recently
from
9669420
to
3a28029
Compare
Signed-off-by: Andrei Kvapil <kvapss@gmail.com>
| | etcdOperator.readinessProbe.periodSeconds | int | `10` | | | ||
| | etcdOperator.resources.limits.cpu | string | `"500m"` | | | ||
| | etcdOperator.resources.limits.memory | string | `"128Mi"` | | | ||
| | etcdOperator.resources.requests.cpu | string | `"100m"` | | |
There was a problem hiding this comment.
Why is the request CPU 100m? This is quite a lot, I think you should start with 20-40m, but you need to know how much etcd eats at least.
| | etcdOperator.readinessProbe.initialDelaySeconds | int | `5` | | | ||
| | etcdOperator.readinessProbe.periodSeconds | int | `10` | | | ||
| | etcdOperator.resources.limits.cpu | string | `"500m"` | | | ||
| | etcdOperator.resources.limits.memory | string | `"128Mi"` | | |
There was a problem hiding this comment.
The memory limit is very small, an average k8s cluster with 100 pods will force etcd to use up to 300MB per instance. I think it should be increased to 384-512MB
| | kubeRbacProxy.readinessProbe | object | `{}` | | | ||
| | kubeRbacProxy.resources.limits.cpu | string | `"500m"` | | | ||
| | kubeRbacProxy.resources.limits.memory | string | `"128Mi"` | | | ||
| | kubeRbacProxy.resources.requests.cpu | string | `"100m"` | | |
There was a problem hiding this comment.
This is quite a lot, I think you should start with 20-40m.
| | kubeRbacProxy.livenessProbe | object | `{}` | | | ||
| | kubeRbacProxy.readinessProbe | object | `{}` | | | ||
| | kubeRbacProxy.resources.limits.cpu | string | `"500m"` | | | ||
| | kubeRbacProxy.resources.limits.memory | string | `"128Mi"` | | |
There was a problem hiding this comment.
It's hard to say here, but it's worth reviewing the limit, you don't really want to catch OOM
|
@themoriarti, there are defaults provided by the etcd-operator maintainers. We can consider other values, but I think it might be better to address this in the upstream project. |
|
We are waiting for a response from etcd-operator aenix-io/etcd-operator#199 , if they make changes it will simplify the integration. |
| resources: | ||
| limits: | ||
| cpu: 500m | ||
| memory: 128Mi |
There was a problem hiding this comment.
need more memory up to 512MB
| cpu: 500m | ||
| memory: 128Mi | ||
| requests: | ||
| cpu: 100m |
There was a problem hiding this comment.
for start etcd not need more then 40m CPU reservation
|
@kvaps Can you add the default value of requests and resource limits when creating an etсd cluster? |
This change replaces kamaji-etcd with our own etcd-operator, which provides a robust environment for running etcd clusters.
The previous kamaji-etcd has not been removed yet, so it will continue to operate for compatibility purposes.
fixes: #33