Question: Keep creating new connections every 5min when using connection pool

[joycezhou0621]

Question: Keep creating new connections every 5min when using connection pool

[mp911de]

Spring Data Redis hosts multiple connection pools. There are separate pools for reactive and for synchronous usage. Also, if your application uses the Cluster API to run commands on specific nodes, then there's a third pool being created.

Please also keep in mind that Cluster connections consist of multiple connections to the individual Redis Cluster nodes.

If you observe a regular disconnect after 5 minutes, then it seems that some active network component disconnects idle connections. Please watch out for Lettuce logs mentioning disconnects and connection watchdog attempting reconnects.

[spring-projects-issues]

If you would like us to look at this issue, please provide the requested information. If the information is not provided within the next 7 days this issue will be closed.

[joycezhou0621]

@mp911de Thank you for your assistance. I have an additional question. I experimented with not using a connection pool and instead utilizing a shared native connection for our redis cluster, while also enabling eager initialization. Here is our setup.

connectionFactory.setShareNativeConnection(true);
connectionFactory.setEagerInitialization(true);
connectionFactory.afterPropertiesSet();

I noticed some logs indicating that a shared connection was created during bean initialization. However, when a new EC2 node starts up and begins serving traffic, we observe a significant number of Redis command timeout errors. Interestingly, these errors only occur for the first few requests, and afterward, they no longer appear.
Due to the at least 5-minute gap between bean initialization and the node actually starting to serve traffic, is it possible that the shared native connection closed during this time interval? However, we have not set any timeout from the server side. Do you have any insights into why these peculiar Redis command timeout errors only occur during node startup?

[mp911de]

Due to the at least 5-minute gap between bean initialization and the node actually starting to serve traffic, is it possible that the shared native connection closed during this time interval?

If Redis is configured to drop a connection, then yes, this might happen. Watch closely the debug log to determine what is happening on the transport level.