Lettuce 4.4.4 Final version During failover the client make a lot of connections to redis

[s-aravind-flipkart]

Redis cluster went into limbo state when the client started to make a lot of connections to redis cluster. I've not enabled any pooling as such. During the failover state the clients were making 1000 connections to 78 boxes.

I think definitely there is a connection leak in the client. loadViews() was called from the client. Redis cluster was not able to handle so many connections and was constantly rebalancing between master to slave, slave to master as such.

How do we avoid this condition in future.

[s-aravind-flipkart]

xxx warning redis[xxxx]: Failover auth denied to 2e07xxx: already voted for epoch 225

[mp911de]

Potentially related tickets: #342, #721.

You can force a connection leak by imposing a lower command execution timeout than the connect timeout (e.g. RedisURI timeout is 10 seconds, timeout set via RedisClusterClient.setDefaultTimeout(…) is higher than 10 seconds). Can you check your timeout settings?

[s-aravind-flipkart]

@mp911de The connection time out is 2x the request timeout. Im trying deploying the config changes to see the effect

[s-aravind-flipkart]

@mp911de Im using the future.get() .get timeout was not same as connection time out.

[s-aravind-flipkart]

The default timeout was 5s in cluster client. In Redis future from lettuce, we used the timeout to 1s.

[mp911de]

Did you change timeouts and is the issue now solved?

[mp911de]

Closing due to lack of requested feedback. If you would like us to look at this issue, please provide the requested information and we will re-open the issue.