sqlldr 報錯：Field in data file exceeds maximum length 解決方法

問題場景

今天有個需求，將 hive 跑完的數據入到 Oracle 里，需求側直接從 Oracle 表里取數據，不接收文件；

 

通過下面腳本落到本地文件：

[hive@hadoop101 tool]$ more hive_2_file.sh
#!/bin/sh

path=$1
query=$2
file=$3
field=$4

beeline -u 'jdbc:hive2://110.110.1.110:9000' -n username -p password -e "insert overwrite directory '/warehouse/servpath/downlo
ad/${path}'
ROW FORMAT SERDE 'org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe'
WITH SERDEPROPERTIES (
'field.delim'='${field}',
'serialization.format'= '',
'serialization.null.format'=''
)
${query}
"
hadoop fs -getmerge /warehouse/servpath/downlo/${path} ${file}

文件入庫控制文件：

load data
characterset utf8
infile '/warehouse/servpath/downlo/TEMP_INFO_20200909.txt'
append into table TEMP_INFO_20200909
fields terminated by "\t"
TRAILING NULLCOLS
(MONTH_ID, PROV_ID, USER_ID, COUNTRY_STAY_DAYS)

執行入庫程序：

#!/bin/sh

table=$1

sqlldr scott/tieger@orcl control=/TEMP_INFO_20200909/${table}.ctl log=/TEMP_INFO_20200909/${table}.log bad=TEMP_INFO_20200909/${table}.bad rows=1100000000 direct=true skip_index_maintenance=TRUE

結果居然報錯了 

#!/bin/shRecord 5031: Rejected - Error on table TEMP_INFO_20200909, column COUNTRY_STAY_DAYS. Field in data file exceeds maximum length 

可是 文件里的字段沒有超過表里的最大長度，在網上找了一下，說是需要修改控制文件，在報錯字段后 + char(300) 即可；

load data
characterset utf8
infile '/warehouse/servpath/downlo/TEMP_INFO_20200909.txt'
append into table TEMP_INFO_20200909
fields terminated by "\t"
TRAILING NULLCOLS
(MONTH_ID, PROV_ID, USER_ID, COUNTRY_STAY_DAYS char(300))

重啟啟動入庫程序，問題得以解決