Problem connecting to Clickhouse

[chungst]

I am trying to fetch data from Clickhouse and pipe them into Polars and I got stuck in the dispatcher.run part.
I have tried both TextProtocol and BinaryProtocol on MySQLSource but that didn't make a difference. The table is show small which only has 98 rows and it shouldn't stuck at Start writing forever. Did I miss anything?

fn run() -> Result<(), Box<dyn Error>> { let database_url = ""; let pool_size = 8; let source = MySQLSource::<TextProtocol>::new(&database_url, pool_size)?; let mut destination = Arrow2Destination::new(); let queries = &["select * from tables"]; let dispatcher = Dispatcher::<MySQLSource<TextProtocol>, Arrow2Destination, MySQLArrow2Transport<TextProtocol>>::new(source, &mut destination, queries, None); let r = dispatcher.run().expect("run failed"); Ok(()) } 2024-03-26T04:35:04.285101Z DEBUG connectorx::dispatcher: Run dispatcher 2024-03-26T04:35:04.285135Z DEBUG connectorx::dispatcher: Prepare 2024-03-26T04:35:04.285154Z DEBUG connectorx::dispatcher: Fetching metadata 2024-03-26T04:35:04.295518Z DEBUG connectorx::dispatcher: Do not need counts in advance 2024-03-26T04:35:04.299392Z DEBUG connectorx::dispatcher: Allocate destination memory: 0x0 2024-03-26T04:35:04.299445Z DEBUG connectorx::dispatcher: Create destination partition 2024-03-26T04:35:04.299459Z DEBUG connectorx::dispatcher: Start writing 

[wangxiaoying]

Hi @chungst , according to the log (Allocate destination memory: 0x0), there is 0 columns in the result, which is pretty weird. Basically 0x0 means # row x # col. # row is 0 for arrow destination since we do not need to preallocate the buffer, but # col should not be 0. Not sure whether it is the reason that caused the issue. Can you make sure that the mysql driver can access your clickhouse?

[wangxiaoying]

Thanks for the investigation!

This is weird, since if the stmt returns with error, our currently logic will seek to generate a query like "SELECT * from branches limit 1" to directly query the database for schema like here. And if it still fails, it should throw an error here with debug and warning messages. However, it seems like there is no such messages printed in your log.

Can you please confirm the code didn't reach to this path?

[chungst]

The code did not reach https://github.com/sfu-db/connector-x/blob/main/connectorx/src/sources/mysql/mod.rs#L126
Given the output from dbg! on these two lines.

[connectorx/src/sources/mysql/mod.rs:102] &first_query = Naked( "select * from branches", ) [connectorx/src/sources/mysql/mod.rs:103] conn.prep(first_query) = Ok( Statement { inner: InnerStmt { columns: None, params: None, stmt_packet: StmtPacket { status: ConstU8( PhantomData<mysql_common::packets::InvalidStmtPacketStatus>, ), statement_id: 0, num_columns: 0, num_params: 0, __skip: Skip, warning_count: 0, }, connection_id: 563, }, named_params: None, }, )

The return of conn.prep is Ok and there is no way it will reach the Err part. Although It looks like the problem started all the way back in the MySQL crate(I could be wrong), I am not sure if there should be a guard internally in connector-x on the Statement

[luong-komorebi]

@chungst I wonder: did you ever get around this problem ? I am working on the same set of dst_arrow, src_mysql and I am also using TextProtocol as well, but i never got it working and the log also shows the same

[2025-03-21T19:41:07Z DEBUG connectorx::get_arrow ] Protocol: text [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Run dispatcher [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Prepare [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Fetching metadata [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Do not need counts in advance [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Allocate destination memory: 0x0 [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Create destination partition [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Start writing [2025-03-21T19:41:07Z DEBUG connectorx::dispatcher ] Finalize partition 0 [2025-03-21T19:41:07Z ERROR <our program> ] Failed to execute query: Invalid argument error: must either specify a row count or at least one column 

[luong-komorebi]

the query that I am running is pretty simple, just a show table or select 1 or select .... (anything)

[chungst]

@luong-komorebi Unfortunately I didn't find a solution and I have moved on.