Bug 2014526

Summary: resolv.conf on Cloud images are not empty, causing cloud-init automatic network configuration to fail
Product: [Fedora] Fedora Reporter: Neal Gompa <ngompa13>
Component: spin-kickstartsAssignee: Neal Gompa <ngompa13>
Status: CLOSED CURRENTRELEASE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 35CC: admiller, awilliam, bruno, dustymabe, kevin, mhayden, ngompa13, robatino, vanmeeuwen+fedora, vpavlin
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard: AcceptedBlocker
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2021-10-18 12:56:53 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 1891955    

Description Neal Gompa 2021-10-15 13:38:47 UTC
Description of problem:
During the kickstart of the cloud image, NetworkManager writes an
/etc/resolv.conf that contains nameserver 192.168.122.1. This causes
boot delays with cloud-init since it does some early boot DNS
redirection tests before talking to the cloud's metadata service. On
some clouds/architectures, this delay is 15 seconds or more.

To fix this, the /etc/resolv.conf needs to be truncated so it can be replaced properly by NetworkManager and cloud-init on the first boot.

Additional info:
This was fixed by Major Hayden in Rawhide in https://pagure.io/fedora-kickstarts/pull-request/846.

Comment 1 Fedora Blocker Bugs Application 2021-10-15 13:41:20 UTC
Proposed as a Blocker for 35-final by Fedora user ngompa using the blocker tracking app because:

 This can cause cloud images to fail to successfully configure and boot in some clouds.

This is also kind of retroactively being filed because I cherry-picked it already without realizing I needed to do this: https://pagure.io/fedora-kickstarts/c/32b03e0440a8717277ebfbb51606bd92f328b54c?branch=f35

Comment 2 Adam Williamson 2021-10-17 20:13:52 UTC
+3 in https://pagure.io/fedora-qa/blocker-review/issue/552 , marking accepted. Do we actually need to do anything here now, or can we just close it?

Comment 3 Neal Gompa 2021-10-18 01:01:11 UTC
I assume at this point we've got a Cloud image build that Major can test with to verify the fix before closing it.

Comment 4 Major Hayden 🤠 2021-10-18 12:00:00 UTC
I'll check the latest images from koji at Vexxhost.

Comment 5 Major Hayden 🤠 2021-10-18 12:37:05 UTC
x86 looks good. The pre-metadata part of cloud-init was taking 10-15 seconds before and it's now 1 second. The whole cloud-init run was 30-45 seconds and is now 18 seconds.

Checking aarch64 now.

Comment 6 Major Hayden 🤠 2021-10-18 12:42:38 UTC
aarch64 is down to 1 second for the pre-metadata part and the whole run is under 45 seconds.

I'm pretty comfortable saying we've fixed this one.

Comment 7 Neal Gompa 2021-10-18 12:56:53 UTC
Excellent, closing it now!