BE crash FailureSignalHandler

[jeason]

【详述】问题详细描述 be有一个节点crash
【背景】做过哪些操作？
【业务影响】
【StarRocks版本】例如：2.4.5
【集群规模】例如：3fe（1 follower+2observer）+9be（fe与be分开）
【机器信息】CPU虚拟核/内存/网卡，例如：32C/128G/万兆
【联系方式】为了在解决问题过程中能及时联系到您获取一些日志信息，请补充下您的联系方式，例如：社区群4-小李或者邮箱，谢谢
【附件】

• fe.log/beINFO/相应截图
• 慢查询：
  • Profile信息
  • 并行度：show variables like ‘%parallel_fragment_exec_instance_num%’;
  • pipeline是否开启：show variables like ‘%pipeline%’;
  • be节点cpu和内存使用率截图
• 查询报错：
  • query_dump，怎么获取query_dump文件
• be crash
  • be.out
    19:15 宕机
    be.out
    query_id:c40e6af1-1412-11ee-876f-5254005f1696, fragment_instance:c40e6af1-1412-11ee-876f-5254005f169e
    *** Aborted at 1687778132 (unix time) try “date -d @1687778132” if you are using GNU date ***
    PC: @ 0x287abaf starrocks::Expr::is_constant()
    *** SIGSEGV (@0x0) received by PID 2285963 (TID 0x7fe7575fc700) from PID 0; stack trace: ***
    @ 0x4946402 google::(anonymous namespace)::FailureSignalHandler()
    @ 0x7fe82494f3ab os::Linux::chained_handler()
    @ 0x7fe824953efc JVM_handle_linux_signal
    @ 0x7fe824946d48 signalHandler()
    @ 0x7fe82401a420 (unknown)
    @ 0x287abaf starrocks::Expr::is_constant()
    @ 0x287abb2 starrocks::Expr::is_constant()
    @ 0x1fee26a starrocks::vectorized::OlapScanConjunctsManager::eval_const_conjuncts()
    @ 0x2252cb7 starrocks::pipeline::OlapScanContext::parse_conjuncts()
    @ 0x2250f49 starrocks::pipeline::OlapScanPrepareOperator::pull_chunk()
    @ 0x1eaa553 starrocks::pipeline::PipelineDriver::process()
    @ 0x3d51a6c starrocks::pipeline::GlobalDriverExecutor:: worker_thread()
    @ 0x380e8b2 starrocks::ThreadPool::dispatch_thread()
    @ 0x38093da starrocks::supervise thread()
    @ 0x7fe82400e609 start_thread
    @ 0x7fe823dd4133 clone
    @ 0x0 (unknown)

[jeason]

对应的query_id和sql排查过，应该没有问题

select time, sum(item_count) item_count, count(shop_id) shop_count, sum(sales) sales, sum(sold) sold, case when sum(sold) != 0 then sum(sales)/sum(sold) else 0 end as avg_price from (select t1.time, count(distinct t1.item_id) item_count, t1.shop_id, sum(t1.sales) sales, sum(t1.sold) sold from (select t1.time, t1.item_id, t1.shop_id, coalesce(max(c1.sales), max(c2.sales), max(t1.sales)) sales, max(cp_sold(date_format(t1.time,’%Y-%m-%d’), c1.sold, c1.sold_cp, c1.sold_min, c1.sold_min_v2, c2.sold, c2.sold_cp, c2.sold_min, c2.sold_min_v2, t1.sold, ‘’, ‘item’, ‘sold’)) sold from item2 t1 left JOIN item2_patch_1687749794 c1 ON t1.cat1 = c1.cat1 AND t1.time = c1.time AND t1.item_id = c1.item_id AND c1.username = ‘游仁Yoren’ and (t1.cat2 = c1.cat2) left JOIN item2_patch_1687749794 c2 ON t1.cat1 = c2.cat1 AND t1.time = c2.time AND t1.item_id = c2.item_id AND c2.username is null and (t1.cat2 = c2.cat2) where ‘游仁Yoren’ = ‘游仁Yoren’ and ((((t1.cat1=‘50010788’ and t1.cat2=‘122438001’) ))) and (t1.time in (‘2023-01-01’, ‘2023-02-01’, ‘2023-03-01’, ‘2023-04-01’, ‘2023-05-01’)) and ((t1.shop_type = ‘cshop’ ) or (t1.shop_type = ‘tmall’ and t1.un1 = 0 )) and (replace(lower(t1.title), ’ ‘,’’) like ‘%silcot%’ or replace(lower(t1.title), ’ ‘,’’) like ‘%舒蔻%’ or replace(lower(t1.title), ’ ‘,’’) like ‘%尤妮佳%’) group by t1.time, t1.shop_id, t1.item_id) t1 group by t1.shop_id, t1.time) t1 group by t1.time

[许秀不许秀]

query dump 提供一下？

[jeason]

dump_file (245.9 KB)

[jeason]

单独执行这个sql也没有出现crash的情况

[jeason]

[jeason]

补充宕机时间点的监控数据