多集群时member集群经常显示未就绪

[ftsxd]

主集群节点上的tower pod报错如下，网络均没有任何问题，但是经常出现，等待2分钟左右又恢复了，
E0115 01:46:14.038945 1 upgradeaware.go:367] Error proxying data from client to backend: read tcp 10.233.78.199:16614->10.233.78.138:48506: read: connection reset by peer

E0115 01:46:14.040117 1 upgradeaware.go:367] Error proxying data from client to backend: read tcp 10.233.78.199:16614->10.233.78.138:48472: read: connection reset by peer

[HuangGuSheng]

一样的问题，ksv3.3.1，tower v0.2.0。1 host + 14member