Skip to content

fix: add retry logic in uefi_setup for hosts with existing BIOS password#39

Merged
parmani-nv merged 1 commit into
mainfrom
fix/bmc_retry_setpassword
Jan 17, 2026
Merged

fix: add retry logic in uefi_setup for hosts with existing BIOS password#39
parmani-nv merged 1 commit into
mainfrom
fix/bmc_retry_setpassword

Conversation

@parmani-nv
Copy link
Copy Markdown
Contributor

Description

When setting the UEFI password on hosts during re-ingestion, the first attempt assumes no password is set (empty current password). This fails if the BMC already has a password set, causing hosts to get stuck in the UefiSetup state.

This can happen when:

  • Force-delete failed to clear the password (network/BMC issues)
  • Host was force-deleted before UEFI clearing was implemented
  • Manual BIOS password was set outside of Forge
  • Previous ingestion was interrupted after password was set

Change
The fix adds retry logic: if the first attempt with empty password fails, retry using the site default password as the current password. This handles re-ingested hosts that still have the site password set on the BMC.

The retry with change(site_pw, site_pw) is effectively a no-op that validates the password matches what Forge expects and allows ingestion to proceed.

Fix
NVBugs 5792728

https://nvbugspro.nvidia.com/bug/5792728
https://nvbugspro.nvidia.com/bug/5122539

@parmani-nv parmani-nv requested a review from a team as a code owner January 16, 2026 19:10
@copy-pr-bot
Copy link
Copy Markdown

copy-pr-bot Bot commented Jan 16, 2026

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

@parmani-nv parmani-nv force-pushed the fix/bmc_retry_setpassword branch from 68516ed to a1203fa Compare January 16, 2026 19:10
When setting the UEFI password on hosts during re-ingestion, the first
attempt assumes no password is set (empty current password). This fails
if the BMC already has a password set, causing hosts to get stuck in
the UefiSetup state.

This can happen when:
- Force-delete failed to clear the password (network/BMC issues)
- Host was force-deleted before UEFI clearing was implemented
- Manual BIOS password was set outside of Forge
- Previous ingestion was interrupted after password was set

The fix adds retry logic: if the first attempt with empty password fails,
retry using the site default password as the current password. This
handles re-ingested hosts that still have the site password set on the BMC.

The retry with change(site_pw, site_pw) is effectively a no-op that
validates the password matches what Forge expects and allows ingestion
to proceed.

Fixes: NVBugs 5792728
Signed-off-by: Parham Armani <parmani@nvidia.com>
@parmani-nv parmani-nv force-pushed the fix/bmc_retry_setpassword branch from a1203fa to 428721d Compare January 17, 2026 00:42
@parmani-nv parmani-nv merged commit 96a312f into main Jan 17, 2026
1 check passed
@ajf ajf deleted the fix/bmc_retry_setpassword branch January 23, 2026 04:01
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants