#975424 cinnamon: systemd NetworkManager units disabled, dhclient still asking for ip at boot

Package:
cinnamon
Source:
cinnamon
Description:
Innovative and comfortable desktop
Submitter:
frankenDist
Date:
2026-08-16 11:57:03 UTC
Severity:
normal
#975424#5
Date:
2020-11-22 00:48:44 UTC
From:
To:
Dear Maintainer,

*** Reporter, please consider answering these questions, where appropriate ***

   * What led up to the situation?
			· I'm trying to disable dhclient, i see this when capturing traffic:
	-----------------------------------
					sudo tcpdump -vv -i enp37s0 |grep -i dhc -a20
	tcpdump: listening on enp37s0, link-type EN10MB (Ethernet), capture size 262144 bytes
	[...]
	01:27:52.961625 IP (tos 0x0, ttl 64, id 0, offset 0, flags [none], proto UDP (17), length 314)
	    0.0.0.0.bootpc > 255.255.255.255.bootps: [udp sum ok] BOOTP/DHCP, Request from [edited_mac_from_my_cablemodem] (oui Unknown), length 286, xid 0x5afcab45, secs 60504, Flags [none] (0x0000)
		  Client-Ethernet-Address [edited_mac_from_my_cablemodem] (oui Unknown)
		  Vendor-rfc1048 Extensions
		    Magic Cookie 0x63825363
		    DHCP-Message Option 53, length 1: Discover
		    Client-ID Option 61, length 7: ether [edited_mac_from_my_cablemodem]
		    Requested-IP Option 50, length 4: 192.168.0.10
		    MSZ Option 57, length 2: 576
		    Parameter-Request Option 55, length 7:
		      Subnet-Mask, Default-Gateway, Domain-Name-Server, Hostname
		      Domain-Name, BR, NTP
		    Vendor-Class Option 60, length 12: "udhcp 1.19.3"
	------------------------------------
		I am almost 100% sure that the MAC address is from my cablemodem, since it only differs in the last bit from the MAC of the cablemodem my Debian OS knows in the arp -a output, the tcpdump mac ends with 2 and the arp -a entry ends with 0.


   * What exactly did you do (or not do) that was effective (or
     ineffective)?
			· The DHCP server in my cablemodem is disabled, the APs are also disabled (just to be clear, the DHCPDISCOVER isn't from other device), no more devices are connected to my main (and only one network device).
			+ I have systemctl disable && systemctl mask the following units:
				- NetworkManager.service
				- NetworkManager-wait-online.service
				- NetworkManager-dispatcher.service
			· I have also edited the line "ExecStart=/usr/lib/NetworkManager/nm-dispatcher" to "ExecStart=" file /etc/systemd/system/dbus-org.freedesktop.nm-dispatcher (which was a symlink to /usr/lib/NetworkManager/nm-dispatcher , and since the mask the symlink doesn't exists any more, now the original NetworkManager*.service are in this path).
			· My /etc/network/interfaces has all the lines commented
			· My /etc/network/interfaces.d/enp37s0 is setted as inet static, i guess allow hotplug isn't involved in this behavior
			· Since i have masked the services ps is not able to find a process:
				ps lax |grep -i dhc |grep -v grep
					ps lax |grep -i dhc |grep -v grep
5 65534    1813       1  20   0  13480  2088 -      S    ?          0:00 /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/vm-nat.conf --leasefile-ro --dhcp-script=/usr/lib/libvirt/libvirt_leaseshelper
1     0    1814    1813  20   0  13480   384 -      S    ?          0:00 /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/vm-nat.conf --leasefile-ro --dhcp-script=/usr/lib/libvirt/libvirt_leaseshelper

			· I have been giving headache to some people in the #debian irc channel, and they have been really patience and helpfull, before masking the service we were able to track with the PID that the resposable for this behavior was ifup..:
				systemctl status ifup@enp37s0.service
				● ifup@enp37s0.service - ifup for enp37s0
					[...]
					Nov 21 23:44:28 deb10x64lpm dhclient[1059]: DHCPDISCOVER on enp37s0 to 255.255.255.255 port 67
					Nov 21 23:44:39 deb10x64lpm dhclient[1059]: No DHCPOFFERS received.
					Nov 21 23:44:39 deb10x64lpm dhclient[1059]: No working leases in persistent database - sleepin
					Nov 21 23:47:23 deb10x64lpm dhclient[1059]: DHCPDISCOVER on enp37s0 to 255.255.255.255 port 67 (*4 times)
					[...]
					Nov 21 23:48:24 deb10x64lpm dhclient[1059]: No DHCPOFFERS received.
					Nov 21 23:48:24 deb10x64lpm dhclient[1059]: No working leases in persistent database - sleepi

				..after the service masking the output isn't the same:
				systemctl status ifup@enp37s0.service
				● ifup@enp37s0.service - ifup for enp37s0
				    Loaded: loaded (/lib/systemd/system/ifup@.service; static)
				    Active: active (exited) since Sun 2020-11-22 00:41:06 CET; 59min ago
	    		Process: 1099 ExecStart=/bin/sh -ec ifup --allow=hotplug enp37s0; ifquery --state enp37s0 (code=exited, status=0/SUCCESS)
				 Main PID: 1099 (code=exited, status=0/SUCCESS)

					Nov 22 00:41:06 deb10x64lpm systemd[1]: Started ifup for enp37s0.
					Nov 22 00:41:12 deb10x64lpm sh[1969]: enp37s0=enp37s0

			· I have tried also a reboot, no luck, same behavior.

   * What was the outcome of this action?
			Still seeing those bootpc packets with tcpdump.

   * What outcome did you expect instead?
			Stop that traffic.

*** End of the template - remove these template lines ***

#975424#10
Date:
2026-08-16 11:56:01 UTC
From:
To:
The tcpdump output attached to the report identifies the sender:

   Vendor-Class Option 60, length 12: "udhcp 1.19.3"

Debian has never shipped that DHCP client. Debian systems use dhclient
(isc-dhcp-client) or NetworkManager's internal DHCP client, neither of
which announces itself as "udhcp", and BusyBox - which provides udhcpc -
has never been at 1.19.3 in any Debian suite: buster had 1:1.30.1-4,
bullseye 1:1.30.1-6, stretch and jessie 1.22.0, wheezy 1.20.0, squeeze
1.17.1. BusyBox 1.19.x dates back to 2011 and is typical of embedded
router firmware.

This matches the reporter's own observation that the source MAC address
belongs to his cable modem, and it explains why the packets kept showing
up after NetworkManager had been masked and no dhclient process was
running any more: the requests come from the modem itself, on the wire,
not from the Debian machine.

The dhclient seen before masking belonged to ifup@enp37s0.service
(ifupdown), so at most that part was a local ifupdown/NetworkManager
configuration question; no Cinnamon component performs DHCP.

Since there has been no follow-up since November 2020 and both the
reported version (cinnamon 4.6.6-1) and the suite (buster) are long out
of the archive, I am closing this report. If DHCP requests really do
originate from a current Debian system against the user's intention,
please open a new bug against the package actually responsible
(ifupdown or network-manager), with "ps" output showing the client
process and the relevant /etc/network/interfaces* configuration.