Starrocks v3.1.3 物化视图刷新导致BE挂掉重启

💬 StarRocks 用户问答 功能使用相关
#1

为了更快的定位您的问题，请提供以下信息，谢谢
【详述】Starrocks v3.1.3 物化视图刷新失败，报com.starrocks.common.UserException: [E110]Fail to read from Socket{id=679 fd=605 addr=10.218.2.226:8060:63256} (0x0x7f99154fed40): Connection timed out [R1][E112]Not connected to 10.218.2.226:8060 yet, server_id=679 [R2][E112]Not connected to 10.218.2.226:8060 yet, server_id=679 [R3][E112]Not connected to 10.218.2.226:8060 yet, server_id=679 错误，经过排查，发现10.218.2.226 这台BE的状态是DEAD
【背景】我们在v3.1.3 进行物化视图异步刷新大数据量压力测试中，出现了一个词BE服务挂掉重启的情况
【业务影响】
【StarRocks版本】例如：1.18.2
【集群规模】例如：3fe（1 follower+2observer）+8be
【机器信息】CPU虚拟核/内存/网卡，例如：48C/64G/万兆
【附件】
select * from information_schema.task_runs where query_id = ‘e1b6d607-64b8-11ee-a15b-bc97e157f93e’

image
image600×564 30.3 KB

226_be_10_26-11_05.log (645.6 KB)

#2

物化视图的创建语句还有base表的数据量请您发下

#3

CREATE MATERIALIZED VIEW cdp_event_model_tangtest0929 (code, event_time, superid, mainid, mainid_type, etl_task_id, dt, ts, created_at, updated_at, task_id, event, touch_point, event_body, _brand) PARTITION BY (date_trunc(‘month’, event_time)) DISTRIBUTED BY HASH(superid) BUCKETS 8 REFRESH MANUAL PROPERTIES ( “replicated_storage” = “true”, “replication_num” = “3”, “storage_medium” = “HDD” ) AS SELECT cdp_event_model.code, cdp_event_model.event_time, cdp_event_model.superid, cdp_event_model.mainid, cdp_event_model.mainid_type, cdp_event_model.etl_task_id, cdp_event_model.dt, cdp_event_model.ts, cdp_event_model.created_at, cdp_event_model.updated_at, cdp_event_model.task_id, cdp_event_model.event, cdp_event_model.touch_point, cdp_event_model.event_body, cdp_event_model._brand FROM cdp_demo.cdp_event_model WHERE cdp_event_model.event = ‘tangtest0929’;

base 表的数据量为（5.9亿）592000049；符合where条件的是（0.66亿）66000000

#4

@dongquan BE只出现了一个挂掉重启的情况，之后报了 ava.lang.RuntimeException: create partitions failed: Table creation timed out. You can increase the timeout by increasing the config “tablet_create_timeout_second” and try again. To increase the config “tablet_create_timeout_second” (currently 10), run the following command: admin set frontend config("tablet_create_timeout_second"="20") or add the following configuration to the fe.conf file and restart the process: tablet_create_timeout_second=20 调整到tablet_create_timeout_second =60之后就没有报错信息，是否和之前tablet_create_timeout_second为默认值有关

#5

有新的排查进展吗？

#6

我们的存储全是SSD的，在创建物化视图没有显式指定“storage_medium” = “HDD” 被默认添加了“storage_medium” = “HDD” 是否影响？这是BE的配置

image
image738×21 2.29 KB

#7

@trueeyu @dongquan

#8

这个问题没被解决吗？
我也是SSD也出现同样问题，怎么解决的 求解

京ICP备2022026421号-1