Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: 启动tikv-server报错
tikv-server data loss cannot start, how to handle it
【TiDB Environment】Test environment
【TiDB Version】5.3.2
【Problem Encountered】[“failed to start node: Engine(Other("[components/raftstore/src/store/fsm/store.rs:1004]: \"[components/raftstore/src/store/peer_storage.rs:482]: [region 2] entry at apply index 278772 doesn’t exist, may lose data.\""))”]
【Reproduction Path】Directly killed the tikv-server service
【Problem Phenomenon and Impact】tikv-server startup error
You can refer to the practice of a fellow member
You can also refer to this
Is a single-node deployment usable?
Can a single-node deployment be used?
The same applies to deploying multiple TiKV instances on a single machine.
Could you please provide some guidance via WeChat or similar? It’s quite urgent. The KV just won’t start no matter what.
Single-node deployment is not unusable because it can be deployed, but it depends on who is using it. It should not be used in a formal testing environment because some behaviors might be strange.
As for your lost data, could it be due to unconfigured disk parameters, a sudden power outage in the virtual machine data center, or issues occurring between storage layers?
The simplest and most direct method is unsafe recovery, as described in the documentation steps provided by the previous experts, but you must accept the possibility of data loss.
Single-machine deployment is possible, but single-node deployment is not feasible; at least three nodes are required.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.
|
username: TiDBer_bGjah8qy
|
username: 
|
username: 
|
username: 
|
username: