ReadTimeout: Error from server: code=1200 [Coordinator node timed out waiting for replica nodes’ responses] message=“Operation timed out for guitardb.active_line_status_details - received only 0 responses from 1 CL=ONE.” info={‘received_responses’: 0, ‘required_responses’: 1, ‘consistency’: ‘ONE’}
how can I resolve this?
This is not enough information to go on. Please check the logs and setup monitoring.
Also, please post cluster topology.
I see you also posted on slack, where you mention this is a full scan query. Those do often time out on larger tables. I will paste here the answer from slack for posterity:
Well you are clearly running a full table scan, so that can easily timeout indeed. This is a bad query, so it is no surprise it is timing out.
This is how you should be doing full table scans Efficient full table scans with ScyllaDB 1.6 - ScyllaDB
An alternative is to raise the timeouts. You can do that with per query timeout, see https://docs.scylladb.com/stable/cql/cql-extensions.html#using-timeout.
This allows raising the timeout for just for a single query, without affecting others. Note that you might also have to adjust your driver’s timeouts as well.
select sum(unit_price) AS line_total from active_line_status_new WHERE order_createts>=‘2022-10-01 00:00:00’ AND order_createts<=‘2022-10-27 23:59:59’ ALLOW FILTERING;
here I have created secondary index on order_createts column.Still Im facing this issue.
The fact that you had to add ALLOW FILTERING signals that the query is still a full scan behind the scenes and faces the same challenges as a full scan.
SELECT queries that have a constrain on a column value like col >= x AND col <= y are always filtering queries, unless col is a clustering key component (and x and y are key prefixes).
Creating a secondary index on order_states will create a material view, where order_states is the partition key. A SELECT query then still has to filter, because partitions are not ordered according to their values, but according to their token.
Hi @Botond_Denes!
Though Im using this query select sum(unit_price) AS line_total from active_line_status_new WHERE order_createts>=‘2022-10-01 00:00:00’ AND order_createts<=‘2022-10-27 23:59:59’ ALLOW FILTERING; , still im facing same issue.
here index is created on order_createts column.
I recommend increasing the range scan timeouts, seeing that you cannot get around a scanning query.
This can be changed by the range_request_timeout_in_ms config item in scylla.yaml. Currently this requires a restart of the node to take effect. Make sure you also adjust timeouts on the client side accordingly.