mobc connection pool

[byblakeorriver]

Hello.
I am trying to track down why I am receiving these errors in my services only after many hours have passed:

Input/output error: Input/output error: Driver error: `Connection to the server is closed.

Here is my setup, and I am running in rust version 1.78.0

diesel = { version = "2.2.2", features = ["chrono"] }
diesel-async = { version = "0.5.0", features=["mysql", "mobc"] }
mobc = "0.8.4"

Here is how I am using it:

let mut manager_config = ManagerConfig::default();
manager_config.recycling_method = RecyclingMethod::Fast;
let manager = AsyncDieselConnectionManager::<AsyncMysqlConnection>::new_with_config(
  DATABASE_URL.clone(),
  manager_config,
);
let pool = Pool::builder()
  .max_open(*DB_POOL_MAX_SIZE)
  .max_idle(*DB_POOL_MAX_IDLE)
  .max_idle_lifetime(Some(Duration::from_secs(
    *DB_POOL_MAX_IDLE_LIFETIME_SECONDS,
  )))
  .test_on_check_out(*DB_POOL_TEST_ON_CHECKOUT)
  .health_check_interval(Some(Duration::from_secs(
    *DB_POOL_HEALTH_CHECK_INTERVAL_SECONDS,
  )))
  .build(manager);
let pool = Arc::new(pool);

I then store that Arc in a context that gets shared throughout my service.

I guess my questions are:

1. Has anyone else seen this behavior?
2. Do you know if this a mobc thing, or a diesel-async while using mobc thing?
3. If it is a diesel-async thing, are there any tools that I can use to help debug this so that I can take a look.

If you need more info please let me know.

[weiznich]

I have not seen this behavior yet, so I'm not sure where it's coming from. I cannot really say what's the best way to debug this.

That written: You configure your pool to use RecyclingMethod::Fast, which essentially skips a check whether the connection is still usable. You might want to prefer using RecyclingMethod::Verified (the default), which performs a check query before assuming that the connection still works. In your particular case it might be that the connection is just closed by the server or by any network interruption. RecyclingMethod::Verified would detect this case and open a new connection instead.

[byblakeorriver]

Good call!

I will try that and report back.

Thank you!

[byblakeorriver]

I have switched to using RecyclingMethod::Verified

After a few hours I was seeing the same error so I don't think it is that.
I am going to continue to try and dig. I will let you know what I find.

[byblakeorriver]

I think it has to do with another bit of my configuration. I had test_on_checkout set to false.
I think that was preventing the checks from happening even with the recycling_method set to Verified.
Going to test my hypothesis and will update once I know for sure.

[byblakeorriver]

Ok, I did finally get around this issue and it was entirely a configuration issue on my end.
Once that was fixed, I did end up getting the same errors in this issue: #139

I am going to use the work around you mentioned for now, and try to get a PR in to use the ping he suggested.
Thank you for your help!