Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: tikvclient_backoff_seconds_count
[TiDB Usage Environment] Production Environment / Testing / Poc
[TiDB Version]
[Encountered Issues]
[Reproduction Path] What operations were performed that led to the issue
[Issue Phenomenon and Impact]
What is the reason for TiDB continuously reporting backoff errors?
[Attachment]
Please provide the version information of each component, such as cdc/tikv, which can be obtained by executing cdc version/tikv-server --version.
Can someone help me take a look at what this error is?
After the TiDB process starts, the Cache is empty. When querying the Cache through Key (or Key Range), it first queries the Cache. If there is a Cache Hit, it directly returns the result from the Cache. If a Miss occurs, it retrieves the involved Region information from PD through Key (Key Range) and backfills it into the Cache.
Once the information is filled into the Cache, it needs to be cleared in the following three situations:
The current implementation of region clearing in TiDB uses mark-and-sweep, i.e., marking for deletion. When the next query arrives and sees the deletion mark, it will treat it as non-existent, triggering a Cache Miss. (This implementation is mainly to reduce lock contention and ensure simplicity and reliability.)
(1) The Region has no access within 10 minutes.
(2) TiKV indicates that the Region information has changed.
(3) The TiKV node provided by the Region is unreachable.
Sorry, I can’t assist with that.
|
username: tidb狂热爱好者
|
username: