Extremely high rate of connection attempts in HA scenario

[jdeppe-pivotal]

We're developing a Redis-compatible cluster backend and are using Lettuce and Spring for some integration testing. One of the test scenarios involves crashing and restarting a backend server while operations are occurring. I've observed the following (after enabling lettuce debug logging):

The thread, attempting to perform an operation against the crashed server, logs over 25000 messages like the following in just 4 seconds:

[debug 2022/04/11 07:08:53.096 PDT redisClientgemfire1_host1_20534 <vm_1_thr_3_redisClient1_host1_20534> tid=0xdd] dispatching command AsyncCommand [type=GET, output=ValueOutput [output=null, error='null'], commandType=io.lettuce.core.protocol.Command]
[debug 2022/04/11 07:08:53.096 PDT redisClientgemfire1_host1_20534 <vm_1_thr_3_redisClient1_host1_20534> tid=0xdd] getConnection(READ, 5466)
[debug 2022/04/11 07:08:53.096 PDT redisClientgemfire1_host1_20534 <vm_1_thr_3_redisClient1_host1_20534> tid=0xdd] [channel=0xf4d67f66, /10.98.40.60:45338 -> /10.98.40.60:20674, epid=0x8] writeToDisconnectedBuffer() Completing command ClusterCommand [command
=AsyncCommand [type=GET, output=ValueOutput [output=null, error='null'], commandType=io.lettuce.core.protocol.Command], redirections=0, maxRedirections=1000] due to connection error

After this massive burst of messages, nothing else is logged by this thread and the operation eventually times out after 1 minute.

Could someone explain what is happening here? Specifically in regards to the huge number of retries in such a short timeframe.