存算分离集群 compaction异常

【详述】broker_load后自动触发compaction,发现compaction失败
【是否存算分离】是
【StarRocks版本】3.1.7
【集群规模】3FE+16CN。
【机器信息】K8S部署+ceph存储

broker_load成功后自动compaction ,但是compaction失败了,查询性能比未compaction前还差。

上面compactions中看到2024-03-26 09:35:40 开始compaction,到2024-03-26 12:51:16发生异常。A error occurred: errorCode=2001 errorMessage:Channel inactive error! 后面应该是系统自动尝试compaction了4次,每次都是Unable to validate object。

表结构
CREATE TABLE u2u_0324 (
source_id bigint(20) NOT NULL COMMENT “”,
target_id bigint(20) NOT NULL COMMENT “”,
equal_gmt_first varchar(64) NULL COMMENT “”,
equal_gmt_last varchar(64) NULL COMMENT “”,
equal_cnt varchar(64) NULL COMMENT “”,
equal_medium_cnt varchar(64) NULL COMMENT “”,
equal_days varchar(64) NULL COMMENT “”,
main_medium varchar(10240) NULL COMMENT “”,
score_all double NULL COMMENT “”,
score_a double NULL COMMENT “”,
score_b double NULL COMMENT “”,
score_c double NULL COMMENT “”,
score_d double NULL COMMENT “”,
score_e double NULL COMMENT “”,
score_f double NULL COMMENT “”,
score_g double NULL COMMENT “”,
score_h double NULL COMMENT “”,
score_i double NULL COMMENT “”,
score_j double NULL COMMENT “”,
equal_event_trace varchar(64) NULL COMMENT “”,
max_sub_gmt_last date NULL COMMENT “”,
ds date NULL COMMENT “”,
score_level tinyint(4) NULL COMMENT “”,
INDEX idx_max_sub_gmt_last (max_sub_gmt_last) USING BITMAP,
INDEX idx_source_id (source_id) USING BITMAP,
INDEX idx_target_id (target_id) USING BITMAP
) ENGINE=OLAP
DUPLICATE KEY(source_id, target_id)
COMMENT “OLAP”
DISTRIBUTED BY HASH(source_id, target_id)
PROPERTIES (
“replication_num” = “1”,
“datacache.enable” = “true”,
“storage_volume” = “builtin_storage_volume”,
“enable_async_write_back” = “false”,
“enable_persistent_index” = “false”,
“compression” = “LZ4”
);

broker_load

LOAD LABEL x_relation.u2u
(
DATA INFILE(“oss://xx/x/export//”)
INTO TABLE u2u_0324
COLUMNS TERMINATED BY “,”
FORMAT AS “csv”
(
enclose = “”"
)
(source_id, target_id, equal_gmt_first, equal_gmt_last, equal_cnt, equal_medium_cnt, equal_days, main_medium, score_all, score_a, score_b, score_c, score_d, score_e, score_f, score_g, score_h, score_i, score_j, max_sub_gmt_last, ds)
set (score_level = floor(score_all*10))
)
WITH BROKER
(
“fs.oss.accessKeyId” = “”,
“fs.oss.accessKeySecret” = “”,
“fs.oss.endpoint” = “”
)
PROPERTIES
(
“timeout” = “86400”
);

目前一个TXID长时间运行未完成 也未失败,不知道啥进度

@trueeyu 大佬可以帮忙确认下,是不是和这个是一样的问题https://forum.mirrorship.cn/t/topic/3928/124#:~:text=%E4%B8%BB%E9%94%AE%E6%A8%A1%E5%9E%8B%E8%A1%A8%20backup/restore%20%E5%90%8E%2C%20BE%20crash%EF%BC%8C%E5%B9%B6%E4%B8%94%E6%97%A0%E6%B3%95%E9%87%8D%E5%90%AF

什么现像,是Crash了?

可以参考文章 https://zhuanlan.zhihu.com/p/688693205