Assorted fixes for terraform-related issues#74
Merged
chucklever merged 4 commits intomainfrom Dec 20, 2025
Merged
Conversation
Cloud VMs sometimes experience intermittent DHCP failures after reboot where DHCP clients do not receive responses even with aggressive retransmission. This issue has been observed primarily on Azure, but could occur on other cloud providers. It typically happens on subsequent reboots, while the first boot during bringup is reliable. Add a new Ansible role dhclient_cache that implements persistent DHCP lease caching with automatic fallback across multiple DHCP mechanisms. The solution detects the DHCP client in use and deploys the appropriate hooks or scripts to save successful DHCP configuration to a persistent cache and automatically restore it if DHCP fails after a reboot. Supported DHCP mechanisms: - ISC dhclient: Used by Debian, Ubuntu, older RHEL/CentOS/Fedora Uses dhclient enter/exit hooks in /etc/dhcp/dhclient-*-hooks.d/ - NetworkManager: Used by RHEL 8+, Fedora 30+, CentOS 8+ Uses dispatcher scripts in /etc/NetworkManager/dispatcher.d/ - wicked: Used by SUSE Linux Enterprise Server, openSUSE Uses wicked extensions in /etc/wicked/extensions/ The cache is valid for one hour and includes IP address, subnet mask, gateway, DNS servers, and other essential network parameters. When DHCP times out or fails, the appropriate mechanism applies the cached configuration using native commands and updates resolv.conf with cached DNS servers. The role is automatically applied during terraform bringup after wait_for_connection succeeds and activates for all terraform-based cloud deployments. The implementation detects the distribution and DHCP mechanism at runtime and configures the appropriate caching solution. Generated-by: Claude AI Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
The libaio1t64 package was introduced as part of Debian's 64-bit time transition starting in Debian 13 (Trixie). Earlier Debian releases use the libaio1 package name instead. The existing logic to override the package name for older releases checked for buster and bookworm but missed bullseye (Debian 11). This caused fstests build dependencies to fail on Debian 11 systems with "No package matching 'libaio1t64' is available". Add bullseye to the list of releases that use the libaio1 package name. Generated-by: Claude AI Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Users need to be able to override the Infrastructure as Code tool binary path when it's installed in non-default locations. The hidden TERRAFORM_BINARY_PATH option made this difficult because it wasn't visible in the configuration interface and used hard-coded paths that don't match typical package manager installations. This change promotes the option to a visible menu item with improved defaults that match common installation paths. The Terraform default now uses /usr/bin/terraform instead of /usr/local/bin/terraform, aligning with standard package manager behavior. The new help text guides users on when and how to override the path for custom installations. Generated-by: Claude AI Signed-off-by: Chuck Lever <cel@kernel.org>
The Ansible template terraform.tfvars.j2 fails with an undefined variable error when generating Azure terraform configuration: AnsibleUndefinedVariable: 'terraform_azure_accelerated_networking_enabled' is undefined This occurs because the Kconfig variable TERRAFORM_AZURE_ACCELERATED_NETWORKING_ENABLED is a hidden bool (no user prompt) that uses conditional defaults based on VM size. When a VM size like Standard_B2s is selected, the variable defaults to 'n' via the condition "default n if TERRAFORM_AZURE_VM_SIZE_STANDARD_B2S". Kconfig's "output yaml" mechanism only writes variables to the yaml output file when they are explicitly set in .config or when hidden bools evaluate to 'y'. Hidden bools that evaluate to 'n' through conditional defaults are never written to the yaml output, leaving the Ansible variable undefined at template render time. The kernel-builder defconfig works because it explicitly sets the variable to 'y', which causes Kconfig to write it to the yaml. The nfsd-fstests defconfig relies on the conditional default, which evaluates to 'n' and is never output. Add a fallback default of 'false' in the gen_tfvars role defaults, following the existing pattern used for AWS variables. This ensures the variable is always defined regardless of whether Kconfig outputs it. Generated-by: Claude AI Signed-off-by: Chuck Lever <cel@kernel.org>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Make instance rebooting more reliable by caching the DHCP-assigned IP address
Support Debian Bullseye when checking for libaio packages
Fix the default path for the "terraform" executable
Work around an "output yaml" quirk