Hi !

I use Heroku PostgreSQL as datasource. Redash is hosted on Docker. I’ve already made a lot of queries and visualisations which worked perfectly.

However, when I execute a complex query (multiple join, case when…), it runs for 5-6min and crash with this error: Error running query: SSL SYSCALL error: EOF detected

According to Heroku documentation, i’ve put sslmode=require in data source: Heroku Postgres | Heroku Dev Center

Thank you for your help,

Saphia

This isn’t really a Redash issue. I searched for that error message and it seems like your database ran out of memory. Try increasing the memory for your Heroku database.

Thank you for your answer.

On Heroku side, it’s written that These errors indicate a client side violation of the wire protocol.

I am wondering if it has any link with my Docker implementation ?

How much resources have you assigned to your docker container? Is it possible your Docker instance ran out of memory?

It was one of my first guess also. I’ve already tried to run the query on Redash while checking docker container memory with docker stats -a. The container memory was ok.

What happens if you run the exact same query from a different query tool? Does it return a similar error?

The exact query worked on Dbeaver connected to the Postgres Heroku Database. It took 9min of run. I think that for long query run, redash closes the connection.

Redash doesn’t close connections arbitrarily. But it’s good to know that the query works from a different place. Please try provisioning more memory for you Docker compose configuration. Specifically the concern here is your worker process.

Thank you @k4s1m for your help.
I have already tried to add memory to the worker through Docker compose configuration but it didn’t work…

We bought a server and since i have installed Redash on it, the query works ! I don’t know what was the problem with my Docker configuration locally (maybe the memory…?) but everything works well now :slight_smile:

1 Like