long running query failure due to client timeout

[Maxim Ulanovskiy]

Hello,

I have 2 long running queries (SELECT INSERT and POPULATE MATERIALIZED VIEW) which fail with client side error after ~45 min:

↙ Progress: 598.86 million rows, 114.01 GB (220.13 thousand rows/s., 41.91 MB/s.) ██████████████████████████████████████████████████                                                                     42%
Exception on client:
Code: 32. DB::Exception: Attempt to read after eof: while receiving packet from 172.18.xx.xx:9000

Connecting to 172.18.xx.xx:9000.
Code: 210. DB::NetException: Connection refused: (172.18.xx.xx:9000)

I thought of network error and rerun query on the same machine where clickhouse-server sits, however result was the same.

I didn't find errors in clickhouse-server logs afterwards.

Queries were killed on the server side though as VIEW POPULATE progress has stopped in the middle.

1. Is this expected behavior of the clickhouse-client?
2. Is there some error log file for clickhouse-client somewhere?
3. Is there some configurable parameter for client/server side timeout which could explain this error?

Thanks,

Max

[Maxim Ulanovskiy]

I reduced number of columns in SELECT and it fixed the problem.

Which means that root cause of the issue was server side failure.

Not sure if Clickhouse had an issue with specific column values or size of data.

In both cases there should have been some traces in the error log I guess but there were none.

[phhcer]

It happened to me also.

Is there any progress?