jepsen 测试遇到的一些问题

OpenAtomFoundation / pikiwidb

a high-performance, large-capacity, multi-tenant, data-persistent, strong data consistency based on raft, Redis-compatible elastic KV data storage system based on RocksDB

BSD 3-Clause "New" or "Revised" License

195 stars 62 forks source link

jepsen 测试遇到的一些问题 #414

Open happy-v587 opened 1 month ago

happy-v587 commented 1 month ago

使用 jepsen 框架测试raft集群时遇到的一些问题：

[ ] 需要支持 multi exec：测试workload需要使用这个功能
[ ] raft join支持hostname方式：目前支持ip。jepsen框架使用hostname做多容器处理（以后k8s部署也会使用hostname这种方式）
[ ] 支持并发 confchange：jepsen 会同时启动4个容器做 raft.cluster join，目前会出现部分节点join失败
[ ] 集群版访问follower节点报错 ERR MOVED 172.18.0.5:9221 client 无法自动处理

Issues-translate-bot commented 1 month ago

Bot detected the issue body's language is not English, translate it automatically.

Title: jepsen testing

Some problems encountered when testing raft clusters using the jepsen framework:

[ ] Need to support multi exec: test workload needs to use this function
[ ] raft join supports hostname mode: currently supports ip. The jepsen framework uses hostname for multi-container processing (k8s deployment will also use hostname in the future)
[ ] Support concurrent confchange: jepsen will start 4 containers at the same time for raft.cluster join. Currently, some nodes will fail to join.
[ ] The cluster version reports an error when accessing the follower node: ERR MOVED 172.18.0.5:9221 The client cannot automatically handle some problems encountered by

dingxiaoshuai123 commented 1 month ago

第四个错误好像是因为写了线性一致性读，然后 follower 无法处理读请求，会返回转移错误。这个稍后需要添加配置文件可开关。

Issues-translate-bot commented 1 month ago

Bot detected the issue body's language is not English, translate it automatically.

The fourth error seems to be caused by writing a linear consistency read, and then the follower cannot process the read request and will return a transfer error. This will need to be added to the configuration file later to switch it on and off.