TiDB Log Error

This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: tidb日志报错

| username: jiangh is the TiDB node address, and is the ospflvs subnet address. LVS is in DR mode. After the cluster was created and LVS was configured, this warning log appeared. Other versions of the TiDB cluster did not have this issue.
[2023/06/25 19:14:38.537 +08:00] [WARN] [server.go:614] [“Server.onConn handshake”] [conn=9127171366494143125] [error=“read tcp> read: connection reset by peer”] [“remote addr”=]

| username: xfworld | Original post link

It seems to be a network issue… the connection was reset…

| username: zhanggame1 | Original post link

Check the network configuration and see if the parameters have been adjusted according to the manual.

| username: redgame | Original post link, check it out.

| username: Ming | Original post link

If other clusters are not configured with LVS, this might be a common issue with load balancing interacting with TiDB. It could be caused by enabling TCP checks, which establish a connection with the database but do not communicate.

You can try disabling the checks first to confirm if this is the reason. If it is, you can change the checks to point to TiDB’s HTTP API to avoid log spamming.

| username: jiangh | Original post link

How to disable the detection?

| username: jiangh | Original post link

Are you referring to LVS detection or TiDB’s?

| username: Ming | Original post link

The lvs, which is the health check mechanism,
As a load balancer, it definitely has this function. I’m not sure how to disable it specifically, but you can look it up. There should be relevant configurations for it.