为什么我使用这个 cql 导入 csv 数据失败？

Question

我尝试使用包含 25000 行的 csv 文件导入关系数据，然后我使用这个密码：

USING PERIODIC COMMIT 500 LOAD CSV WITH HEADERS FROM 'http://127.0.0.1:88/static/relations/261.csv' AS line  
MATCH (a:user{value:line.userId}),(b:device{value:line.deviceId})  
MERGE (a)-[rel:myrelname]->(b)  
WITH rel, COALESCE(rel.eventOccurTime, []) + line.eventOccurTime AS eventOccurTime
UNWIND eventOccurTime as r0
WITH rel, collect(distinct r0) AS unique
set rel.eventOccurTime= unique ,rel.business_name='data-test'

30 分钟后，我收到此错误： [Neo.DatabaseError.Statement.ExecutionFailed]：连接重置，参考 0b8a3600-d715-44bf-8c5b-4260edeab459.

我该如何解决？

Answer 1

一些尝试：

简化（并加快）查询。无需展开然后 re-collect 列表的不同元素。除非确实需要，否则无需创建临时列表或附加到列表。

USING PERIODIC COMMIT 500 LOAD CSV WITH HEADERS FROM 'http://127.0.0.1:88/static/relations/261.csv' AS line  
MATCH (a:user{value: line.userId}), (b:device{value: line.deviceId})  
MERGE (a)-[rel:myrelname]->(b)
SET
  rel.business_name='data-test',
  rel.eventOccurTime = CASE
    WHEN rel.eventOccurTime IS NULL THEN [line.eventOccurTime]
    WHEN line.eventOccurTime IN rel.eventOccurTime THEN rel.eventOccurTime
    ELSE rel.eventOccurTime + line.eventOccurTime END

您可能希望通过在以下位置创建 indexes 来提高性能：
- :user(value)
- :device(value)
使用PROFILE分析从你的Cypher生成的执行计划。

为什么我使用这个 cql 导入 csv 数据失败？

why I use this cql import csv data failed?

cql

neo4j

cypher