I'm not familiar enough with ruby to know how to discover/narrow down
the class of the exception - it doesn't appear to be part of the stack
trace - so I'm rescuing very broadly, and retrying. Since there's a
retry limit, in the case where ssh is really unreachable this should
still terminate.
So far instead of seeing as much as a 50% (false) failure rate, I haven't seen any failures with this patch.
This is because we're still running into the same issue as described at https://github.com/mitchellh/vagrant-aws/issues/74#issuecomment-54758045, more than 2 years after it was intially described.
I'm not familiar enough with ruby to know how to discover/narrow down the class of the exception - it doesn't appear to be part of the stack trace - so I'm rescuing very broadly, and retrying. Since there's a retry limit, in the case where ssh is really unreachable this should still terminate.
So far instead of seeing as much as a 50% (false) failure rate, I haven't seen any failures with this patch.