Timeout during long queries when SSH tunnel enabled.

[Onlinehead]

Describe the bug Timeout during a long database queries.

Logs

2023-04-07 14:28:18 
   :card-name "*********",

2023-04-07 14:28:18 
   :card-id 7441,

2023-04-07 14:28:18 
   :context :dashboard,

2023-04-07 14:28:18 
  {:executed-by 119,

2023-04-07 14:28:18 
  :info

2023-04-07 14:28:18 
  :middleware {:js-int-to-string? true, :ignore-cached-results? false},

2023-04-07 14:28:18 
  :type :query,

2023-04-07 14:28:18 
 {:constraints {:max-results 10000, :max-results-bare-rows 2000},

2023-04-07 14:28:18 
 :preprocessed

2023-04-07 14:28:18 
 :running_time 0,

2023-04-07 14:28:18 
 :row_count 0,

2023-04-07 14:28:18 
 "DefaultConnectFuture[ec2-user@/44.196.206.5:22]: Failed to get operation result within specified timeout: 30000",

To Reproduce

1. Run long query over the database which has a SSH tunnel enabled.

Expected behavior Metabase should respect a main database connection timeout.

Screenshots/videos

[paoliniluis]

I just hit this with DataGrip, so I'm guessing we'll have to do some magic here not to hit this (seems it's not our fault)

[camsaul]

@paoliniluis, what is the problem exactly? The SSH tunnel has a 30 second timeout or something, and that's getting triggered while the query is still running?

[paoliniluis]

@camsaul i believe that we’re not checking the health of the tunnel so when it gores down we don’t recreate it. The fact that I hit the same bug with DataGrip puzzles me though

[luizarakaki]

We are closing this for now. We tried repro this with our own data warehouse and couldn't get the same behavior. If new evidence appears, we open this again.