diff --git a/Buildroot/board/FOG/FOS/rootfs_overlay/usr/share/fog/lib/funcs.sh b/Buildroot/board/FOG/FOS/rootfs_overlay/usr/share/fog/lib/funcs.sh
index f9c9ec5..807d682 100644
--- a/Buildroot/board/FOG/FOS/rootfs_overlay/usr/share/fog/lib/funcs.sh
+++ b/Buildroot/board/FOG/FOS/rootfs_overlay/usr/share/fog/lib/funcs.sh
@@ -1440,8 +1440,12 @@ getHardDisk() {
                     matched=1
                     found_match=1
                     disks="$disks $dev"
-                    # remove matched dev from the pool
-                    devs="$(echo " $devs " | sed "s# $dev # #g; s/^ *//; s/ *$//")"
+                    # Remove the matched dev from the pool. $devs is one
+                    # device per LINE, so match whole lines -- a space-padded
+                    # sed never matched, the device stayed in the pool, and
+                    # the "add unmatched devices" step below appended it a
+                    # second time (fogproject #743: /dev/sda captured twice).
+                    devs="$(grep -vx -- "$dev" <<<"$devs")"
                     break
                 fi
             done
diff --git a/tests/checks/primary-disk-dedup.sh b/tests/checks/primary-disk-dedup.sh
new file mode 100755
index 0000000..38aba3c
--- /dev/null
+++ b/tests/checks/primary-disk-dedup.sh
@@ -0,0 +1,98 @@
+#!/bin/bash
+#
+# Assertion harness for getHardDisk() in funcs.sh when Host Primary Disk is
+# set: the named disk goes first and every other disk follows exactly once.
+#
+#   tests/checks/primary-disk-dedup.sh   # run all cases, exit non-zero on any failure
+#
+# fogproject issue #743: with a multi-disk-all image and Host Primary Disk set
+# to /dev/sda, FOS captured /dev/sda, then captured it again as the "next"
+# disk, and on deploy wrote it back twice. getHardDisk() builds the device
+# pool from lsblk one device per LINE, then tried to drop a matched device
+# from the pool with a sed that only matches it with a SPACE on each side.
+# The pattern never matched, nothing was removed, and the final
+# `disks="$disks $devs"` appended the matched disk a second time.
+#
+# What this harness locks:
+#
+#   1. fdrive naming the first-enumerated disk yields each disk once, the
+#      named one first.
+#   2. fdrive naming a later disk moves it to the front and the rest keep
+#      their enumeration order, still once each.
+#   3. A comma-separated fdrive naming two disks yields them in the order
+#      given, with the remainder after.
+#
+# Mechanism mirrors tests/checks/ntfs-shrink-retry.sh: source a sandbox copy
+# of the library and PATH-shadow lsblk, blockdev and blkid with doubles that
+# describe a fixed three-disk machine.
+
+HERE="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_LIB="$HERE/../../Buildroot/board/FOG/FOS/rootfs_overlay/usr/share/fog/lib"
+
+[[ -f $REPO_LIB/funcs.sh ]] || { echo "ERROR: cannot find funcs.sh under $REPO_LIB" >&2; exit 2; }
+
+SANDBOX="$(mktemp -d)"
+trap 'rm -rf "$SANDBOX"' EXIT
+
+cp "$REPO_LIB/partition-funcs.sh" "$SANDBOX/partition-funcs.sh"
+sed -e "s#^\. /usr/share/fog/lib/partition-funcs\.sh#. $SANDBOX/partition-funcs.sh#" \
+    "$REPO_LIB/funcs.sh" > "$SANDBOX/funcs.sh"
+
+STUBBIN="$SANDBOX/bin"
+mkdir -p "$STUBBIN"
+
+# lsblk double. The pool query (-dpno KNAME,SIZE) lists three disks, one per
+# line, the way the real tool does. The per-device SERIAL,WWN query answers
+# with empty values so only the path can match a spec.
+cat > "$STUBBIN/lsblk" <<'STUB'
+#!/bin/bash
+case "$*" in
+    *KNAME,SIZE*)
+        printf '/dev/sda 500G\n/dev/sdb 1T\n/dev/nvme0n1 2T\n'
+        ;;
+    *SERIAL,WWN*)
+        printf 'SERIAL="" WWN=""\n'
+        ;;
+esac
+STUB
+cat > "$STUBBIN/blockdev" <<'STUB'
+#!/bin/bash
+case "$2" in
+    /dev/sda) echo 500107862016 ;;
+    /dev/sdb) echo 1000204886016 ;;
+    *)        echo 2000398934016 ;;
+esac
+STUB
+cat > "$STUBBIN/blkid" <<'STUB'
+#!/bin/bash
+exit 2
+STUB
+chmod +x "$STUBBIN"/*
+export PATH="$STUBBIN:$PATH"
+
+# shellcheck disable=SC1090
+. "$SANDBOX/funcs.sh" >/dev/null 2>&1
+handleError() { echo "handleError: $*" >&2; return 1; }
+
+fail=0
+check() {
+    local name="$1" spec="$2" want="$3"
+    fdrive="$spec"
+    imgType="mpa"
+    type="up"
+    hd=""; disks=""
+    getHardDisk 2>/dev/null
+    if [[ "$disks" == "$want" && "$hd" == "${want%% *}" ]]; then
+        echo "ok    $name"
+    else
+        echo "FAIL  $name: fdrive='$spec' gave disks='$disks' hd='$hd', wanted '$want'"
+        fail=1
+    fi
+}
+
+check "first-enumerated disk named once"   "/dev/sda"          "/dev/sda /dev/sdb /dev/nvme0n1"
+check "later disk moves to the front"      "/dev/sdb"          "/dev/sdb /dev/sda /dev/nvme0n1"
+check "two specs keep their given order"   "/dev/nvme0n1,/dev/sda" "/dev/nvme0n1 /dev/sda /dev/sdb"
+
+[[ $fail -eq 0 ]] && echo "PASS" || echo "FAIL"
+exit $fail