I've commented in on issue, PTAL









--

Best Regards
Mingyu Chen

Email:
morning...@apache.org





At 2025-03-20 11:29:16, "潘庭锋" <pantingf...@pcitech.com> wrote:
>Dear Doris Community/Support Team,
>I hope this email finds you well. I am writing to report an issue encountered 
>while querying a Hive external table via Doris, which leads to BE node 
>crashes. Your expertise would be greatly appreciated in resolving this problem.
>
>
>github issue:&nbsp;&nbsp;https://github.com/apache/doris/issues/49034
>
>
>
>Environment Details
>
>Doris Version:&nbsp;3.0.2
>
>
>Hive Version:&nbsp;3.1.0.3.1.5.0-152
>
>
>
>Issue Description:
>
>When querying a Hive external table with multiple non-partition column filters 
>(e.g.,&nbsp;PERSON_ID,&nbsp;RECORD_ID,&nbsp;CARD_PHYSICAL_ID), all BE nodes 
>crash and return the error&nbsp;io.grpc.StatusRuntimeException: UNAVAILABLE: 
>Network closed for unknown reason.
>
>
>
>
>
>Error Details:
>●Client Error:复制SQL error [1105] [HY000]: RpcException, msg: 
>io.grpc.StatusRuntimeException: UNAVAILABLE: Network closed for unknown reason 
>&nbsp;
>●Reproducibility:
>○Stably Reproducible&nbsp;when querying with non-partition column filters on a 
>large table (~300 million rows).
>○Works Correctly&nbsp;if:
>■Query includes&nbsp;all partition 
>columns&nbsp;(stat_day&nbsp;and&nbsp;order_type).
>■Query is simple (e.g., no&nbsp;WHERE&nbsp;clause or only partition column 
>filters).
>■Table size is small.
>
>
>
>
>
>
>
>
>Request for Support:
>Could you help investigate:
>1.&nbsp;Why non-partition column filters cause BE crashes on large tables?
>2.&nbsp;Whether this is related to metadata propagation, predicate pushdown, 
>or resource constraints (e.g., memory/network)?
>3.&nbsp;Required configurations to avoid such crashes.
>
>
>Thank you for your assistance!
>
>Best regards,
>
>
>Tingfeng Pan&nbsp;

Reply via email to