Skip to content

Conversation

@jphenow
Copy link
Member

@jphenow jphenow commented Nov 11, 2025

WIP

jeromegn and others added 30 commits July 15, 2024 08:20
Note: no longer abort processing if bundle install fails
jphenow and others added 19 commits November 13, 2025 16:10
Fixed two bugs in the updateConfig CPU/Memory override loops:

1. Changed from value-based to index-based iteration
   - Before: for _, c := range (modifies copies, changes lost)
   - After: for i := range (modifies actual slice elements)

2. Added defensive conditionals to only set empty values
   - Prevents conflicts with updateComputeFromDeprecatedGuestFields
   - Only fills in missing CPUKind/CPUs/MemoryMB values

This maintains the compatibility layer for plan-level compute overrides
while avoiding the bugs that were causing preflight test failures.

Fixes preflight test failures in indices 7, 8, 13.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>
The Compute struct has an embedded *fly.MachineGuest pointer, and the
fields CPUKind, CPUs, and MemoryMB are on that embedded pointer. The
recent commit 3817455 fixed the index-based iteration but was still
missing the check for MachineGuest != nil.

Without this check, the code panics with a nil pointer dereference when
trying to access these fields on a nil MachineGuest pointer.

Fixes TestLaunchCpusMem and other related test failures.
@@ -1,26 +1,43 @@
package scanner

import (
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

odd that these changes are here

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

how do you mean

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

idk, felt out of place but I might be wrong

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

lubien and others added 9 commits November 18, 2025 11:13
All plan command functions (runPropose, runCreate, runPostgres, runRedis, runTigris) were ignoring errors returned from RunPlan(), causing commands to exit with status 0 even when they failed.

This caused silent failures where flyctl would fail (e.g., authorization errors) but exit successfully, making debugging extremely difficult.

Fixed by properly returning errors from RunPlan() in all plan command functions.
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.

7 participants