Merge pull request #15824 from Security-Onion-Solutions/fix/reinstall2

fix reinstall issue with salt

setup/so-functions

@@ -1547,8 +1547,6 @@ reinstall_init() {
 		local salt_services=( "salt-minion" )
 	fi
 
-	local service_retry_count=20
-
 	{
 		# remove all of root's cronjobs
 		crontab -r -u root
@@ -1563,31 +1561,48 @@ reinstall_init() {
 
 		salt-call state.apply ca.remove -linfo --local --file-root=../salt
 
-		# Kill any salt processes (safely)
+		# Stop salt services and force-kill any lingering salt processes (including orphans
+		# from an earlier reinstall attempt where the unit file is gone but processes survive)
+		# so dnf remove salt can run cleanly
 		for service in "${salt_services[@]}"; do
-			# Stop the service in the background so we can exit after a certain amount of time
 			if check_service_status "$service"; then
-				systemctl stop "$service" &
+				info "Stopping $service via systemctl"
+				systemctl stop "$service"
 			fi
-			local pid=$!
-
-			local count=0
-			while check_service_status "$service"; do
-				if [[ $count -gt $service_retry_count ]]; then
-					echo "Could not stop $service after 1 minute, exiting setup."
-
-					# Stop the systemctl process trying to kill the service, show user a message, then exit setup
-					kill -9 $pid
-					fail_setup
-				fi
-				
-				sleep 5
-				((count++))
-			done
 		done
 
+		# Unconditionally force-kill any remaining salt binaries — these may be orphaned
+		# from a prior aborted reinstall (no unit file, so systemctl can't see them).
+		for salt_bin in salt-master salt-minion salt-call salt-cloud; do
+			if pgrep -f "/usr/bin/${salt_bin}" > /dev/null 2>&1; then
+				info "Force-killing lingering $salt_bin processes"
+				pkill -9 -ef "/usr/bin/${salt_bin}" 2>/dev/null
+			fi
+		done
+		# Catch stray `salt` CLI children from saltutil.kill_all_jobs / state.apply invocations
+		pkill -9 -ef "/usr/bin/python3 /bin/salt" 2>/dev/null
+
+		# Give the kernel a moment to reap the killed processes before dnf removes the binaries
+		local kill_wait=0
+		while pgrep -f "/usr/bin/salt-" > /dev/null 2>&1; do
+			if [[ $kill_wait -gt 10 ]]; then
+				info "Salt processes still present after SIGKILL + 10s wait; proceeding anyway"
+				pgrep -af "/usr/bin/salt-" | while read -r line; do info "  lingering: $line"; done
+				break
+			fi
+			sleep 1
+			((kill_wait++))
+		done
+
+		# Clear the 'failed' state SIGKILL left on the units before removing the package
+		systemctl reset-failed salt-master.service salt-minion.service 2>/dev/null || true
+
 		# Remove all salt configs
-		rm -rf /etc/salt/engines/* /etc/salt/grains /etc/salt/master /etc/salt/master.d/* /etc/salt/minion /etc/salt/minion.d/* /etc/salt/pki/* /etc/salt/proxy /etc/salt/proxy.d/* /var/cache/salt/
+		dnf -y remove salt
+		rm -rf /etc/salt/ /var/cache/salt/
+
+		# Drop systemd's in-memory references to the now-removed units
+		systemctl daemon-reload
 
 		if command -v docker &> /dev/null; then
 			# Stop and remove all so-* containers so files can be changed with more safety