You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
[oocm] support for externalTrafficPolicy: Local (kubernetes#1720)
* support for externalTrafficPolicy local
with externalTrafficPolicy local we create an http healthmonitor that checks the dedicated healthCheckNodePort.
Signed-off-by: Fabian Ruff <fabian.ruff@sap.com>
* remove limitation from docs
Signed-off-by: Fabian Ruff <fabian.ruff@sap.com>
* address feedback
* Add some note that health monitors are mandatory for externalTrafficPolicy: Local
Defines whether to create health monitor for the load balancer pool, if not specified, use `create-monitor` config. The health monitor can be created or deleted dynamically.
176
+
Defines whether to create health monitor for the load balancer pool, if not specified, use `create-monitor` config. The health monitor can be created or deleted dynamically. A health monitor is required for services with `externalTrafficPolicy: Local`.
177
177
178
178
Not supported when `lb-provider=ovn` is configured in openstack-cloud-controller-manager.
<!-- END doctoc generated TOC please keep comment here to allow auto update -->
22
21
@@ -209,7 +208,7 @@ Although the openstack-cloud-controller-manager was initially implemented with N
209
208
This option is not supported for Octavia. The worker nodes and the Octavia amphorae are usually in the same subnet, so it's sufficient to config the port security group rules manually for worker nodes, to allow the traffic coming from the the subnet IP range to the node port range(i.e. 30000-32767).
210
209
211
210
* `create-monitor`
212
-
Indicates whether or not to create a health monitor for the service load balancer. Default: false
211
+
Indicates whether or not to create a health monitor for the service load balancer. A health monitor required for services that declare `externalTrafficPolicy: Local`. Default: false
213
212
214
213
* `monitor-delay`
215
214
The time, in seconds, between sending probes to members of the load balancer. Default: 5
@@ -289,9 +288,3 @@ Refer to [Metrics for openstack-cloud-controller-manager](../metrics.md)
289
288
### OpenStack availability zone must not contain blank
290
289
291
290
`topology.kubernetes.io/zone` is used to label node and its value comes from availability zone of the node, according to [label spec](https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/#syntax-and-character-set) it does not support blank (' ') but OpenStack availability zone supports blank. So your OpenStack availability zone must not contain blank otherwise it will lead to node that belongs to this availability zone register failure, see [#1379](https://github.com/kubernetes/cloud-provider-openstack/issues/1379) for further information.
292
-
293
-
### externalTrafficPolicy support
294
-
295
-
`externalTrafficPolicy` denotes if this Service desires to route external traffic to node-local or cluster-wide endpoints. "Local" preserves the client source IP and avoids a second hop for LoadBalancer and Nodeport type services, but risks potentially imbalanced traffic spreading. "Cluster" obscures the client source IP and may cause a second hop to another node, but should have good overall load-spreading.
296
-
297
-
openstack-cloud-controller-manager only supports `externalTrafficPolicy: Cluster` for now.
0 commit comments