Monitor free disk percentage, not just absolute space

[yacovm]

Why this should be merged

We currently have metrics for available disk space and we fail the health check if we have less than a predefined amount of space left.

This commit adds metrics for available disk percentage, and fails the health check if we have less than a predefined percentage of space left.

The motivation is that if one wants to re-use the configuration of the validator to several nodes with different volume sizes, using absolute space is less fitting than using percentages.

For example, a volume that started with all of its 2TB free and now has 200GB space left is not the same as than a volume that started with 400GB and has half of its space left.

How this works

Adds a flag --system-tracker-disk-warning-threshold-available-space-percentage that defaults to 3% space left.

Changes AvailableBytes on storage_unix.go, storage_openbsd.go, to return also percentage, not only absolute.

How this was tested

Only tested on Linux, however I looked at the Statfs_t struct of OpenBSD to make sure I picked the right fields.

I took a Fuji node I had and compiled the avalanchego binary from this branch, and re-launched the node and then I created a big file using fallocate -l 400G bigfile and observed the metrics:

Soon enough, the node's health check began failing:

curl 'http://localhost:9650/ext/health'  | jq

I then deleted the file and observed the metrics:

And checked that the node returned to be healthy once more:

Need to be documented in RELEASES.md?

[Copilot]

Pull Request Overview

This PR adds disk space percentage monitoring alongside the existing absolute disk space tracking. The system now tracks both available disk bytes and available disk percentage, allowing for more flexible disk space health checks.

Key Changes:

• Modified AvailableBytes() to return both available bytes and percentage of free disk space
• Added new configuration option for percentage-based disk space warning threshold (default 97%)
• Updated health checks to monitor both absolute and percentage-based disk space thresholds

Reviewed Changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
utils/storage/storage_unix.go	Modified to calculate and return percentage of available disk space
utils/resource/usage.go	Added percentage tracking fields and methods to resource manager
utils/resource/resourcemock/user.go	Added mock implementation for percentage method
utils/resource/no_usage.go	Updated no-op implementation return values
snow/networking/tracker/resource_tracker.go	Added percentage tracking to disk resource tracker
node/node.go	Updated health check to validate percentage threshold
config/node/config.go	Added percentage threshold configuration field
config/keys.go	Added configuration key for percentage threshold
config/flags.go	Added command-line flag for percentage threshold
config/config.go	Updated disk space config validation to handle percentage

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[rrazvan1]

Suggested change

	err = fmt.Errorf("remaining available disk space percentage (%d%%) is below minimum required available space percentage (%d%%)", availableDiskPercentage, n.Config.WarningThresholdAvailableDiskSpacePercentage)
	err = fmt.Errorf("remaining available disk space percentage (%d%%) is below the warning threshold of disk space percentage (%d%%)", availableDiskPercentage, n.Config.WarningThresholdAvailableDiskSpacePercentage)

[yacovm]

1634be6