Fix k8s watch #573
Fix k8s watch #573
Conversation
|
Thank you, to clarify, does this mean you only need to label the ingress and then specify that label in traefik config? |
|
Yes that is the idea, the label selector in the traefik config is there in order to select witch ingresses to use, it does not make any sense to use this selector on services and pods since the ingress already defines this selection. |
|
Yep, +1 On 1 August 2016 01:08:24 BST, Ed Robinson notifications@github.com wrote:
Sent from my Android device with K-9 Mail. Please excuse my brevity. |
|
LGTM :) |
emilevauge
added
contributor/needs-resolve-conflicts
and removed
status/1-needs-design-review
labels
|
Ping @emilevauge could we get this merged before we release the next version, or we are going to be breaking released behaviour. |
|
LGTM 🐵 |
|
How can I see when this gets released? |
|
It needs a rebase to get merged 😇 |
We added the ability to filter the ingresses used by traefik based on a label selector, but we shouldn't need to have matching labels on every other resource, Ingress allready has a way to explicty choose which pods end up in the load ballancer (by refering to the membership of a particular service)
|
@vdemeester rebased |
It was mentioned in the support slack that if the pods and services are not labeled with the same label then changes are not picked up by the watch.
This reverts the change to the watch for endpoints and services so should fix the issue.
I do feel that at some point we should refactor the k8s client a bit so we can do many highly scoped watches rather than one per resource type... but this works for now.