How to watch jaeger-agent disconnection

[Asisranjan]

I am using jaeger-all-in-one docker image
When trace registered first time, everything works.
if pod/container restarts the application which generates trace is not saved and shown in Jaeger UI.

Suspecting - when pod/container restart, agent connection lost. (Please help me to correct my understanding if I am wrong)

[yurishkuro]

Which transport are you using to report spans? The default UDP does not keep connection, and expects the agent to run on the local host. However, if you run the agent under a DNS name (even if it's on the same host), then some Jaeger clients resolve DNS name to IP on start-up, so if that IP changes because of agent restart then the clients may "lose the connection". This would need to be addressed in specific clients (there were fixes in some languages, I don't recall which).

[Asisranjan]

Which transport are you using to report spans? The default UDP does not keep connection, and expects the agent to run on the local host.
UDP transport
However, if you run the agent under a DNS name (even if it's on the same host), then some Jaeger clients resolve DNS name to IP on start-up, so if that IP changes because of agent restart then the clients may "lose the connection". This would need to be addressed in specific clients (there were fixes in some languages, I don't recall which).
Using kubernetes service name as Jaeger agent host
Using Jaeger Java client to report span. Is the point i mentioned a real issue? Do you have any open item to address this in java client ?

Can you please suggest the mechanism to solve above problem

[yurishkuro]

There were some fixes in Java and Go clients for this:

• Make UdpSender lazy to be able to recover from early DNS issues jaeger-client-java#726
• Add resolved udp connection type, continually resolve dns names in background jaeger-client-go#520

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[stale]

This issue has been automatically closed due to inactivity.