Something in the Ubuntu 1804 (20200130 update) release broke our tests

[sean-krail]

Describe the bug
Something in the Ubuntu 1804 (20200130 update) release broke our tests. We're not sure what. Starting on 1/31 (20200130 was released 1/30), our smoke tests running in Node.js in docker-compose started throwing ETIMEDOUT errors when making external requests like:

UnhandledPromiseRejectionWarning: RequestError: Error: connect ETIMEDOUT <ip>:<port>

<ip>:<port> being one of our ELBs. So we suspect there's a DNS resolution issue.

We've confirmed this isn't a code change by opening a pull request that reverts back to a commit prior to 1/30, and it still fails with the above error.

We've also tested with Ubuntu 16.04 20200130 and it does not have this issue.

Area for Triage:
Deployment/Release

Question, Bug, or Feature?:
Bug

Virtual environments affected

• macOS 10.15
• Ubuntu 16.04 LTS
• Ubuntu 18.04 LTS
• Windows Server 2016 R2
• Windows Server 2019

Expected behavior
Expect to not see this DNS resolution issue like we did in the 20200119 release. Ideally we could pin to this release using the runs-on property like runs-on: ubuntu-18.04@20200119.

Actual behavior
See the "Describe the bug" section above. I don't have a public repo to share that reproduces this bug.

[sean-krail]

DNS resolution confirmed. We added a step just before the smoke tests to append a line to the /etc/hosts file for the external hostname that we request. Logic looks something like this:

      - name: Update /etc/hosts with External IP address
        run: |
          external_host=$(<ugly, fragile logic to get the hostname dynamically>)
          ip_address="$(dig +short ""$external_host"" | head -n 1)"
          echo "appended line to /etc/hosts: $ip_address $external_host"
          docker exec "$container_making_external_request" \
            sh -c "echo '$ip_address $external_host' >> /etc/hosts"

Smoke tests are passing consistently now, but this is less than ideal.

[al-cheb]

Hello, @sean-krail
The AzP.20200211.ubuntu18.1 has been deployed. Please try to check your test once more time.

[sean-krail]

Hi @al-cheb, it's still failing with the same ETIMEDOUT error

[al-cheb]

Hello, @sean-krail

Could you please replace dns server to 8.8.8.8 and verify the test:

- name: Set 8.8.8.8 as dns server
  run: |
    sudo sed -i 's/#DNS=/DNS=8.8.8.8 8.8.4.4/g' /etc/systemd/resolved.conf
    sudo systemctl daemon-reload
    sudo systemctl restart systemd-networkd
    sudo systemctl restart systemd-resolved

[sean-krail]

That worked for us, @al-cheb!

[hiradp]

@al-cheb - is this something that will be fixed on the virtual environment itself? Or do we need to permanently modify our workflow? CC: @sean-krail

[al-cheb]

Hello, @sean-krail and @hiradp
I will try to escalate the issue and provide you an answer asap. Is it possible to provide dns hostnames that you have faced the issue?

[sean-krail]

@al-cheb, we're no longer able to reproduce the issue with the latest ubuntu build ubuntu18/20200217.1. We never tested ubuntu18/20200211.1 fyi. I'm not sure what changed that would've fixed this issue for us. We'll continue to monitor it on our end.

[alepauly]

Closing this for now since it doesn't repro and we didn't find anything on our side to explain it. @sean-krail if it pops up again please let us know. thank you.