Bug #2786
closedUtilDns test suite fails with some (faulty?) DNS resolver configuration
0%
Description
This may be related to #2653, but I cannot confirm this.
The following are the steps with vagrant that I was able to reproduce the error:
mkdir test
cd test
vagrant up ubuntu/vivid64
vagrant ssh
# install ndn-cxx dependencies
# build ndn-cxx with tests and optimized mode
./build/unit-tests -t UtilDns -l all
Expected:
- test passed
Actual:
- tests failed
There is something funky with default VirtualBox DNS settings. The error goes away when the following is added into vagrant configuration:
config.vm.provider "virtualbox" do |vb|
vb.customize ["modifyvm", :id, "--natdnshostresolver1", "on"]
end
Some related info is here: http://serverfault.com/questions/453185/vagrant-virtualbox-dns-10-0-2-3-not-working
Updated by Davide Pesavento over 9 years ago
What fails exactly, and how? is it a timeout?
Updated by Alex Afanasyev over 9 years ago
I'm still not sure. According to Eric, the failure is caused by the fact that caching resolver does not give IPv6 records when resolving hostname. The solution that was applied on some jenkins bots is adding 8.8.8.8 to the list of caching resolvers.
Updated by Davide Pesavento over 9 years ago
What caching resolver are you using? Do you have control over its configuration?
Btw why don't we hardcode the addresses in /etc/hosts
on the jenkins machines?
Updated by Junxiao Shi over 7 years ago
- Is duplicate of Bug #3270: Occasional non-reproducible test failure in UtilDns/AsynchronousV6 added
Updated by Junxiao Shi over 7 years ago
- Category changed from Utils to Network
- Status changed from New to Duplicate