Update to stop using deprecated grpc/naming package

[bboreham]

It is gone from the latest gRPC release 1.30.

I found a notice announcing this: grpc/grpc-go#3180

Also interesting: why didn't the deprecated package raise a lint warning?

[nicolai86]

I'd like to pick this up. Any objections?

Following the discussion in grpc/grpc#12295, the proposed fix seems to switch to a server side managed connection approach using MaxConnectionAge and MaxConnectionAgeGrace, which will force clients to re-connect defined by MaxConnectionAgeGrace.

As I understand the query frontend code this would break how frontends fan-out query processing to workers today. @bboreham is that a fair statement?

If we want to stick to using polling we have to write our own dns watcher (or pull in the deleted code), or we adjust the interaction between the query frontend and the worker.

[stale]

This issue has been automatically marked as stale because it has not had any activity in the past 60 days. It will be closed in 15 days if no further activity occurs. Thank you for your contributions.

[pracucci]

Still valid. Required by future upgrade to gRPC 1.3.0.

[stale]

This issue has been automatically marked as stale because it has not had any activity in the past 60 days. It will be closed in 15 days if no further activity occurs. Thank you for your contributions.

[jtlisi]

still valid

[stale]

This issue has been automatically marked as stale because it has not had any activity in the past 60 days. It will be closed in 15 days if no further activity occurs. Thank you for your contributions.

[pracucci]

Still valid

[jtlisi]

Unblocked now that etcd-io/etcd#12709 has been merged

[stale]

This issue has been automatically marked as stale because it has not had any activity in the past 60 days. It will be closed in 15 days if no further activity occurs. Thank you for your contributions.

[pracucci]

We're not using the upstream grpc/naming  anymore but, due to convenience, we've imported some of its code in pkg/util/grpc/naming. I'm going to close this issue, given we don't have plans to get rid of it, but feel free to reopen it if you want to reconsider it.