melin

> Calcite is referenced by arrow jdbc. Since arrow jdbc is not used yet, the related dependency will be removed in the next version. commons jar ，spark 中也存在，麻烦去掉，容易导致冲突。

Trino already supports iceberg.query-partition-filter-required config property and query_partition_filter_required session property.

类似hadoop yarn 日志聚合功能，例如spark 运行结果后，如果yarn开启日志聚合，yarn logs applicationId 命令非常方便查看任务运行。运行在k8s，如果任务挂掉，查看日志不方便，需要主动去收集，增加使用成本。

> You means collect the job's logs for it when its log aggregation enabled? yes， k8s is the most troublesome to collect logs, if volcano completes this part of the...

类emr serverless spark 能够自动把日志搜集到s3，任务运行中和结束能够看到日志， 

> I understand that log aggregation should be the capability of the cloud platform rather than the core of the scheduling component. yarn 日志聚合能力，就是调度引擎提供的。这样能力需要调度引擎结合才能实现，平台只是一个展现能力

https://mp.weixin.qq.com/s/az-PIMtGNlxgT5P4KUYmBQ

> Please write issue in English. See https://github.com/StarRocks/starrocks-kubernetes-operator/blob/main/doc/change_root_password_howto.md for more details. Just by setting the environment variables, the initial password was not set. When starting CN, it would keep prompting...

jdk switch to 11, ready to compile. Thanks! jdk8 is used a lot, which is not very convenient，For example, spark antlr4 can only be upgraded to 4.9.3 https://github.com/apache/spark/pull/37814

> > Antlr runtime is compatible with jdk8. Antlr tool requires jdk11. Can you provide a concrete example of this Spark incompatibility (stack trace) ? > > Do you mean...