GreatRiver
GreatRiver
bvt的峰值内存大概11G左右，跑完结束等待几分钟，大概res是8G。load tpch 期间增加4G左右的内存，宿主机15G，docker limit也是15G，如果跑完bvt 内存释放不及时，马上load tpch 1G，mo-service很可能超过14G，这样宿主机就会因为oom kill mo-service。手动测试了一下，跑完bvt 7.7G，load tpch峰值能到11G，load完成后过段时间又回到7.7G。感觉可以增加内存到18G就可以解决，或者调整mem cache为默认的512M，现在是1.5G
[mem.zip](https://github.com/user-attachments/files/21463475/mem.zip) 007 load 前 009 load 后 010 load 后等待几分钟
https://grafana.ci.matrixorigin.cn/explore?panes=%7B%22Ykg%22:%7B%22datasource%22:%22loki%22,%22queries%22:%5B%7B%22refId%22:%22A%22,%22expr%22:%22%7Bnamespace%3D%5C%22mo-branch-nightly-34de694e7-20250609%5C%22,%20pod%3D%5C%22nightly-regression-dis-tp-cn-jw8wf%5C%22%7D%20%7C%3D%20%60slow%20event%60%22,%22queryType%22:%22range%22,%22datasource%22:%7B%22type%22:%22loki%22,%22uid%22:%22loki%22%7D,%22editorMode%22:%22builder%22%7D%5D,%22range%22:%7B%22from%22:%221749529440000%22,%22to%22:%221749529606000%22%7D%7D%7D&schemaVersion=1&orgId=1 在这个时间段有一个cn s3的读压力很大，延迟很高，导致rpc timeout
https://grafana.ci.matrixorigin.cn/d/ae3ttwut16qdca/fileservice-metrics?orgId=1&from=1749529200000&to=1749531599000&var-interval=1m&var-namespace=mo-branch-nightly-34de694e7-20250609&var-pod=nightly-regression-dis-tp-cn-jw8wf
https://github.com/matrixorigin/matrixone/actions/runs/13968520227/job/39104416001?pr=21572
这是tn这边基于GC的checkpoint优化，待1.3 checkpoint merge 完成后再开始开发