Question

我想在cassandra中选择表的特定字段并将它们插入另一个表中。我在sql server中这样做：

INSERT INTO Users(name,family)
SELECT name,family FROM Users

如何在cassandra-cli或cqlsh中使用？

Answer 1

COPY keyspace.columnfamily1 (column1, column2,...) TO 'temp.csv';
COPY keyspace.columnfamily2 (column1, column2,...) FROM 'temp.csv';

这里给出你的键空间（schema-name）而不是columnfamilyname1使用你要复制的表，并在columnfamily2中给出你要复制的表名..

是的，这是CQL从未尝试使用CLI的解决方案..

Answer 2

对于不是很大的表，请保存文件并使用匿名管道：

cqlsh -e "COPY keyspace.src_table (col1, col2, ...,ColN ) TO STDOUT WITH HEADER=false" | cqlsh -e "COPY keyspace.target_table (col1, col2, ...,ColN ) FROM STDIN"

对于非常大的数据集，这不起作用。应探索每个令牌范围的策略

Answer 3

对于非常大的表，CQLSH将很难处理COPY TO / FROM。以下是如何通过本地令牌范围扫描表并将所需列从一个表复制到另一个表。因为它使用本地副本，所以要在数据中心的每个节点上执行：

    #!/bin/bash
    #
    # Script to COPY from a SOURCE table's select columns over a TARGET table
    # Assumes the following:
    #
    # * The SOURCE table is very large - otherwise just try:
    #   cqlsh  -e "COPY keyspace.src_table (col1, col2, ...,ColN ) TO STDOUT WITH HEADER=false" \
    #   |cqlsh -e "COPY keyspace.tgt_table (col1, col2, ...,ColN ) FROM STDIN"
    # * SOURCE AND TARGET TABLES are in the SAME KEYSPACE
    # * TARGET columns are named the SAME as SOURCE
    #
    # The script sweeps thru the local tokens to copy only the local data over to the new table
    # Therefore, this script needs to run on every node on the datacenter
    #
    # Set these variables before executing
    #
    USR=cassandra
    PWD=password
    KSP=my_keyspace
    SRC=src_table
    COL="col1, col2, col3"
    PKY="col1"
    TGT=tgt_table

    CQLSH="cqlsh -u ${USR} -p ${PWD} -k ${KSP}"

    function getTokens(){
       for i in $($CQLSH -e "select tokens from system.local;" | awk -F, '/{/{print $0}' | tr -d '{' | tr -d '}' | tr -d ','); do
           echo ${i//\'/}
       done | sort -n
    }

    function getDataByTokenRange(){
       i=0
       tokens=($(getTokens))
       while [ ${i} -lt ${#tokens[@]} ]; do 
             [ ${i} -eq 0 ]         && echo "SELECT ${COL} FROM ${SRC} WHERE token(${PKY}) <= ${tokens[i]};" 
             [ "${tokens[i+1]}" ]   && echo "SELECT ${COL} FROM ${SRC} WHERE token(${PKY}) >  ${tokens[i]} AND token(${PKY}) <= ${tokens[i+1]};"
             [ ! "${tokens[i+1]}" ] && echo "SELECT ${COL} FROM ${SRC} WHERE token(${PKY}) >  ${tokens[i]};"
             ((i++))
       done
    } 

function cqlExec(){ 
  while IFS='' read -r cql || [[ -n "$line" ]]; do
    $CQLSH -e "CONSISTENCY LOCAL_ONE; $cql"                                                                \
    |awk -F\| '( !/LOCAL_ONE/ && !/'"${COL/, /|}"'/ && !/^\-+/ && !/^\([0-9]+ rows)/ && !/^$/ ){print $0}' \
    |sed -e 's/^[ ]*//g' -e 's/[ ]*|[ ]*/|/g'                                                              \
    |$CQLSH -e "COPY ${TGT} (${COL}) FROM STDIN WITH DELIMITER = '|' and HEADER=false;"
    [ "$?" -gt 0 ] && echo "ERROR: Failed to import data from command: ${command}"
  done < "$1"
}

    main(){
       echo "Begin processing ..."
       getDataByTokenRange > getDataByTokenRange.ddl
       cqlExec getDataByTokenRange.ddl
       echo "End procesing"
    }

    main

如何从表中选择数据并插入另一个表？

3 个答案: