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
http_server_* metric has clientaddress label. earlier client Address has IP and now it is changed to name - while verifying the metric output it matches to our "client pod Deployment name"(not sure that is what Beyla is fetching). but the problem is not able to diagnose to the exact client when there is same name in different namespace connected to single destination.
(for example POD1("monitoring") to POD2 communication here client is also a POD) in k8s environment.
We have a setup like this.
app1 in namespace X is talking to Nginx(pod)
app1 in namespace Y is talking to Nginx (pod)
in http_server metric we have client address label which represents the source of the connection here app1.
problem 1: we are not sure which app1 is connected to Nginx
problem 2: if both are connected to Nginx we will get only one metric for both the entry
Expected to have k8s service name with namespace (app1.ns1.svc.cluster.local) in clientaddress label or existing IP Address is fine. this would define the clear picture of the client from where the connection originated.
The text was updated successfully, but these errors were encountered:
http_server_* metric has clientaddress label. earlier client Address has IP and now it is changed to name - while verifying the metric output it matches to our "client pod Deployment name"(not sure that is what Beyla is fetching). but the problem is not able to diagnose to the exact client when there is same name in different namespace connected to single destination.
(for example POD1("monitoring") to POD2 communication here client is also a POD) in k8s environment.
We have a setup like this.
in http_server metric we have client address label which represents the source of the connection here app1.
problem 1: we are not sure which app1 is connected to Nginx
problem 2: if both are connected to Nginx we will get only one metric for both the entry
Expected to have k8s service name with namespace (app1.ns1.svc.cluster.local) in clientaddress label or existing IP Address is fine. this would define the clear picture of the client from where the connection originated.
The text was updated successfully, but these errors were encountered: