This is not yet implementation, just API and tests to be filled
with implementation in next CLs,
tracked by: https://github.com/etcd-io/etcd/issues/12652
We propose here 3 packages:
- clientv3/naming/endpoints ->
That is abstraction layer over etcd that allows to write, read &
watch Endpoints information. It's independent from GRPC API. It hides
the storage details.
- clientv3/naming/endpoints/internal ->
That contains the grpc's compatible Update class to preserve the
internal JSON mashalling format.
- clientv3/naming/resolver ->
That implements the GRPC resolver API, such that etcd can be
used for connection.Dial in grpc.
Please see the grpc_naming.md document changes & grpcproxy/cluster.go
new integration, to see how the new abstractions work.
Signed-off-by: Chao Chen <chaochn@amazon.com>
Useful for deciding when to terminate the unhealthy follower.
If the follower is receiving a leader snapshot, operator may wait.
Signed-off-by: Gyuho Lee <leegyuho@amazon.com>
This PR adds another probing routine to monitor the connection
for Raft message transports. Previously, we only monitored
snapshot transports.
In our production cluster, we found one TCP connection had >8-sec
latencies to a remote peer, but "etcd_network_peer_round_trip_time_seconds"
metrics shows <1-sec latency distribution, which means etcd server
was not sampling enough while such latency spikes happen
outside of snapshot pipeline connection.
Signed-off-by: Gyuho Lee <leegyuho@amazon.com>
Distribution would be:
0.1 second or more
...
25.6 seconds or more
51.2 seconds or more
etcd_network_snapshot_send_success
etcd_network_snapshot_send_failures
etcd_network_snapshot_send_total_duration_seconds
etcd_network_snapshot_receive_success
etcd_network_snapshot_receive_failures
etcd_network_snapshot_receive_total_duration_seconds
Signed-off-by: Gyuho Lee <leegyuho@amazon.com>
