TravisCI Builds are failing for ppc64le+ruby 2.4

[karlwilbur]

With the introduction of testing on ppc64le (added in PR #69), while building on TravisCI, we are seeing that Bundler will sometimes pull in activerecord > 5, which should not happen. This is not consistent.

Here is an example of failures for both 2.4.7 and 2.4.10
Build number 153: https://travis-ci.com/github/K-and-R/email_validator/builds/222180216

Yet here, with only a change to the README, only 2.4.10 gets activerecord > 5. This seems to be a bug in TravisCI.
Build number 154: https://travis-ci.com/github/K-and-R/email_validator/builds/222703882

I would expect that Bundler behave in a consistent manner across builds for the same Ruby version.

I am going to continue to monitor this. I will also try to find time to research what the cause may be. Right now, I am considering this a low priority since it doesn't impact the actual gem functionality.

(cc: @nageshlop)

[karlwilbur]

Quoting this message from the PR thread:

Hi @karlwilbur
I verified the test by executing travis build on master branch. I was also observing timeout issue, after retrying the build it is successful. Looks intermittent issue in connecting host. Could you rerun the build and check. The build logs are as mentioned below.
Successful build log:
https://travis-ci.com/github/nageshlop/email_validator/builds/222362933
Error:
https://travis-ci.com/github/nageshlop/email_validator/builds/222357740

[karlwilbur]

Now, simply restarting the build of master resulted in a passing build:
Build number 155: https://travis-ci.com/github/K-and-R/email_validator/builds/222706329

[karlwilbur]

I am closing this because I have just removed Ruby 2.4 form the tester version of Ruby.