zetcd errors with etcd master branch

[gyuho]

I have simple local etcd server + zetcd process and tried to send requests to zetcd but almost all requests error...

Here's how to reproduce

# build etcd
./install-zetcd.sh
./start-server-zetcd.sh
./test-zk.sh

https://github.com/gyuho/zookeeper-tests
https://github.com/gyuho/zookeeper-tests/blob/master/start-server-zetcd.sh
https://github.com/gyuho/zookeeper-tests/blob/master/test-zk.sh

[heyitsanthony]

Running the benchmark, I see a lot of timeouts:

2016/12/28 00:24:57 Recv loop terminated: err=read tcp 127.0.0.1:34170->127.0.0.1:2181: i/o timeout
2016/12/28 00:24:57 Send loop terminated: err=<nil>

The zk logs show two things zetcd doesn't implement:

Renegotiating timeouts; benchmark requests 1s, zk gives 4s:

2016-12-28 08:13:09,594 [myid:] - INFO  [SyncThread:0:ZooKeeperServer@645] - Established session 0x159447d61890033 with negotiated timeout 4000 for client /172.17.0.1:51658

Limiting max connections to 60 (but zetcd accepts all 100):

2016-12-28 08:13:10,491 [myid:] - WARN  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:ZooKeeperServer@854] - Connection request from old client /172.17.0.1:51748; will be dropped if server is in r-o mode
2016-12-28 08:13:10,491 [myid:] - INFO  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:ZooKeeperServer@900] - Client attempting to establish new session at /172.17.0.1:51748
2016-12-28 08:13:10,491 [myid:] - WARN  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:NIOServerCnxnFactory@188] - Too many connections from /172.17.0.1 - max is 60

If I change the session timeout to 4s in zetcd like zk does, the time-outs all go away.

Potential fixes would be to support timeout renegotiation (probably worthwhile) and connection limiting (seems silly) like zk.