Maxim

@gahancorpcfo I replaced original host with `somehost.com`, cause original host produces these errors maybe once a month, so there's no reason to try to reproduce the error with it. @sigmavirus24...

@sigmavirus24 yes, you're right in some ways, but I think this particular case should be treaten as connect timeout, cause > Requests that produced this error are [safe to retry](http://docs.python-requests.org/en/master/api/#requests.ConnectTimeout)...

@sigmavirus24 we're using sentry as exception collector with our app, but I cannot share the original issue because our sentry instance is local. But I can explain stacktrace with screenshots...

Now I see clearly, @nateprewitt, you're right, I've posted an issue urllib3/urllib3#1366 with reference to this one

@sigmavirus24 neither SSLError, nor ReadTimeout, I guess. It should be a ConnectionError, since it represents an error, produced while connecting to remote host, and subject request is safe to retry.

@sigmavirus24 I'm sorry if I misled you with `ConnectionError`, I meant `ConnectTimeout`. I think of this situation as of borderline case between connection timeout and read timeout. Intuition acts as...

Sorry for such meager information provided. Singleton tasks isn't retrying at all, seems that task methods is not the best place to hook into. May be celery signals is.

Just looked at signals implementation, and think it'll be a liitle harder than through methods

This issue is actually quite annoying, can we get it handled by django-ninja @vitalik?

I have the same problem too with cyrillic subject