From ef3b393fd4a4481e27f9dd84ca80724e0e8eb651 Mon Sep 17 00:00:00 2001
From: Andrew Durbin <andrewd@zededa.com>
Date: Fri, 19 Apr 2024 12:21:13 -0600
Subject: [PATCH] Check allocated vs provisioned file size in metrics.

Thin allocated/sparse files will report full virtual/max
size in ls but allocated in du.  Comparable stat output
is in %s and %b respectively.  Populate DiskMetric.UsedBytes
with file allocated bytes to avoid rolling over negative when
eve node is populated with persistent volume instances with
little data allocated.  For example an empty system with
a single unattached 1GB persistent volume instance
defined would see the following in
publishMetrics() of handlemetrics.go:

persistUsage = ~1 MB
persistAppUsage = 1 GB
runtimeStorageOverhead = (persistUsage-persistAppUsage)
	= ~17592186042378 MB

ls vs du:
ls -l /persist/clear/volumes/9599c42f-42b0-4807-96a7-2d01cb424e8f\#0.raw
-rw-r--r-- 1 root root 1073741824 Apr 19 17:31 /persist/clear/volumes/...
du -h /persist/clear/volumes/9599c42f-42b0-4807-96a7-2d01cb424e8f\#0.raw
4.0K	/persist/clear/volumes/9599c42f-42b0-4807-96a7-2d01cb424e8f#0.raw

stat virt vs. allocated:
stat -c '%s' /persist/clear/volumes/9599c42f-42b0-4807-96a7-2d01cb424e8f\#0.raw
1073741824
stat -c '%b' /persist/clear/volumes/9599c42f-42b0-4807-96a7-2d01cb424e8f\#0.raw
8

Signed-off-by: Andrew Durbin <andrewd@zededa.com>
(cherry picked from commit 8be1999c60dc5c257a73a7143b096572d3fa89ec)
---
 pkg/pillar/diskmetrics/usage.go | 29 +++++++++++++++++++++++++++--
 1 file changed, 27 insertions(+), 2 deletions(-)

diff --git a/pkg/pillar/diskmetrics/usage.go b/pkg/pillar/diskmetrics/usage.go
index 3b813545c6..12d4b9e993 100644
--- a/pkg/pillar/diskmetrics/usage.go
+++ b/pkg/pillar/diskmetrics/usage.go
@@ -9,6 +9,7 @@ import (
 	"os"
 	"strconv"
 	"strings"
+	"syscall"
 
 	"github.com/containerd/containerd/mount"
 	"github.com/lf-edge/eve/pkg/pillar/base"
@@ -18,6 +19,17 @@ import (
 	"github.com/shirou/gopsutil/disk"
 )
 
+// StatAllocatedBytes returns the allocated size of a file in bytes
+// This value will be less than fileInfo.Size() for thinly allocated files
+func StatAllocatedBytes(path string) (uint64, error) {
+	var stat syscall.Stat_t
+	err := syscall.Stat(path, &stat)
+	if err != nil {
+		return uint64(0), err
+	}
+	return uint64(stat.Blocks * int64(stat.Blksize)), nil
+}
+
 // SizeFromDir performs a du -s equivalent operation.
 // Didn't use ioutil.ReadDir and filepath.Walk because they sort (quick_sort) all files per directory
 // which is an unnecessary costly operation.
@@ -57,8 +69,21 @@ func SizeFromDir(log *base.LogObject, dirname string) (uint64, error) {
 				log.Tracef("Dir %s size %d\n", filename, size)
 				totalUsed += size
 			} else {
-				log.Tracef("File %s Size %d\n", filename, location.Size())
-				totalUsed += uint64(location.Size())
+				// FileInfo.Size() returns the provisioned size
+				// Sparse files will have a smaller allocated size than provisioned
+				// Use full syscall.Stat_t to get the allocated size
+				allocatedBytes, err := StatAllocatedBytes(filename)
+				if err != nil {
+					log.Errorf("StatAllocatedBytes: %s failed %s treating as fully allocated\n", filename, err)
+					allocatedBytes = uint64(location.Size())
+				}
+				// Fully Allocated: don't use allocated bytes
+				// stat math of %b*%B as it will over-account space
+				if allocatedBytes >= uint64(location.Size()) {
+					allocatedBytes = uint64(location.Size())
+				}
+				log.Tracef("File %s Size %d\n", filename, allocatedBytes)
+				totalUsed += allocatedBytes
 			}
 		}
 	}