版本和组织结构：
- doris 版本 2.15; 操作系统版本 ubuntu 20.4
- 结构 1FE + 3BE.
问题描述:
在3个BE的在线使用的情况下，发现简单查询出现900s 查询超时。
初步排查是其中一个BE异常。尝试下线该异常be后，该问题修复。该be上线后问题再次出现。
考虑的配置和环境问题，这边依照正常工作的be做了配置检查，确定系统配置一致。
关键报错日志：
- 以下是故障BE的关键日志。
Runtime Filter Dependency Information:
Finish Dependency Information:
W20241008 11:22:04.951169 4044 pipeline_x_fragment_context.cpp:154] PipelineXFragmentContext cancel instance: 167540fb1174959-8f85fb2740c0b6dc
W20241008 11:22:05.886970 3320 fragment_mgr.cpp:1151] Query 167540fb1174959-8f85fb2740c0b6db is timeout
W20241008 11:22:05.942006 4039 pipeline_x_fragment_context.cpp:147] PipelineXFragmentContext is cancelled due to timeout : PipelineXFragmentContext Info:
Tasks in instance 0:
Task 0: QueryId: 49532e75252d4f76-8ced737c85bf2145
InstanceId: 49532e75252d4f76-8ced737c85bf2148
PipelineTask[this = 0x7fad8fcb2800, state = BLOCKED_FOR_SOURCE, dry run = false, elapse time = 900.004260293s], block dependency = AGGREGATION_OPERATOR_DEPENDENCY: id=6, block task = 1, ready=false, _always_ready=false, is running = false
operators:
AGGREGATION_OPERATOR: id=6, parallel_tasks=1
DATA_STREAM_SINK_OPERATOR: id=7, Sink Buffer: (_should_stop = false, _busy_channels = 0, _is_finishing = false), _reach_limit: false0. AGGREGATION_OPERATOR_DEPENDENCY: id=6, block task = 1, ready=false, _always_ready=false
Write Dependency Information:
ExchangeSinkQueueDependency: id=7, block task = 0, ready=true, _always_ready=false
BroadcastDependency: id=7, block task = 0, ready=true, _always_ready=false
诚心请教doris使用的社区老鸟和各位社区大神。
