0

I was importing about 4 million rows from CSV using COPY. 90 minutes into the process, I got the following error which suggests some problem in waiting for replica nodes. The Keyspace is setup with replication factor of 1 and there is only one node in the cluster. So why does the coordinator have to wait for any other node?

Processed 4050000 rows; Write: 624.27 rows/ss
code=1100 [Coordinator node timed out waiting for replica nodes' responses] message="Operation timed out - received only 0 responses." info=

{'received_responses': 0, 'required_responses': 1, 'write_type': 0, 'consistency': 1}
Aborting import at record #4050617. Previously-inserted values still present.
4050671 rows imported in 1 hour, 26 minutes, and 43.649 seconds.

I have already seen this question: Coordinator node timed out waiting for replica nodes in Cassandra Datastax while insert data

But that does not answer why would a coordinator expect replica nodes to be present when the replication factor is 1 and it is a 1 node cluster.

Some more info:

Node Status

student@cascor:~$ nodetool status
Datacenter: datacenter1
=======================
Status=Up/Down
|/ State=Normal/Leaving/Joining/Moving
--  Address    Load       Tokens  Owns    Host ID                               Rack
UN  127.0.0.1  1.64 GB    256     ?       f550e955-00f3-49a4-bc2d-c7f775079359  rack1

Note: Non-system keyspaces don't have the same replication settings, effective ownership information is meaningless

The Keyspace config

cqlsh:excelsior> desc keyspace;

CREATE KEYSPACE excelsior WITH replication = {'class': 'SimpleStrategy', 'replication_factor': '1'}  AND durable_writes = true;
Community
  • 1
  • 1
Dojo
  • 5,374
  • 4
  • 49
  • 79

1 Answers1

3

Is explain it: 'received_responses': 0, 'required_responses': 1 so u have one node, but it working as u have many. So it can not respond. And it because writing operation need time, when u import so many data in one go it kill CPU and write operation take longer time, so get time out, and node respond "i not save this data" so having 1 node make none respond so u get this error. I had same problem, and changing this (write_request_timeout_in_ms) help me, i to do COPY import of millions of record.

Sędziwój
  • 66
  • 7
  • Thanks for your response. A few days after I posted this question, I started using a tool for loading data into Cassandra insted of the native `COPY` command. But I'll give this a try if I need to use `COPY` again. – Dojo Jan 23 '16 at 14:48
  • If u back, try it, it works great, but u must remember it eat ram memory, so is good but have some weakspot. – Sędziwój Jan 29 '16 at 13:46