I've been running F30 on my Lenovo P72 laptop since upgrading to the beta. I had all sorts of SELinux problems until I reset the contexts. Now there's one last problem I haven't been able to fix and that's the wired Ethernet port. Whenever I connect to it, the NetworkManager applet loops and loops without ever getting a DHCP address. It worked fine in F29, and so does the WiFi port with DHCP in F30.
systemd baffles me, and the lack of /var/log/messages doesn't give me much to look thru. I've run 'journalctl -u NetworkManager' to see what happens when I plug in the RJ-45 connector, then disconnect it:
May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6461] device (enp0s31f6): carrier: link connected May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6464] device (enp0s31f6): state change: unavailable -> disconnected (reason 'carrier-changed', sys-iface-state: 'managed') May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6478] device (enp0s31f6): Activation: starting connection 'Ethernet' (8b81adbf-22bf-3f18-ba5e-85f78dc4fa17) May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6480] device (enp0s31f6): state change: disconnected -> prepare (reason 'none', sys-iface-state: 'managed') May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6484] device (enp0s31f6): state change: prepare -> config (reason 'none', sys-iface-state: 'managed') May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6845] device (enp0s31f6): state change: config -> ip-config (reason 'none', sys-iface-state: 'managed') May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6848] dhcp4 (enp0s31f6): activation: beginning transaction (timeout in 45 seconds) May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6863] dhcp4 (enp0s31f6): dhclient started with pid 21697 May 25 02:18:33 tiger.protogeek.org dhclient[21697]: DHCPDISCOVER on enp0s31f6 to 255.255.255.255 port 67 interval 4 (xid=0x94236275) May 25 02:18:37 tiger.protogeek.org dhclient[21697]: DHCPDISCOVER on enp0s31f6 to 255.255.255.255 port 67 interval 4 (xid=0x94236275) May 25 02:18:41 tiger.protogeek.org dhclient[21697]: DHCPDISCOVER on enp0s31f6 to 255.255.255.255 port 67 interval 11 (xid=0x94236275) May 25 02:18:48 tiger.protogeek.org NetworkManager[1217]: <info> [1558768728.5573] device (enp0s31f6): state change: ip-config -> unavailable (reason 'carrier-changed', sys-iface-state: 'managed') May 25 02:18:48 tiger.protogeek.org NetworkManager[1217]: <info> [1558768728.5740] dhcp4 (enp0s31f6): canceled DHCP transaction, DHCP client pid 21697 May 25 02:18:48 tiger.protogeek.org NetworkManager[1217]: <info> [1558768728.5740] dhcp4 (enp0s31f6): state changed unknown -> done
The wired Ethernet never comes up, but as soon as I unplug it the WiFi port's DHCP connection resumes perfectly.
Can anyone suggest the next step?
--Doc Savage
On 5/25/19 10:32 PM, Robert G. (Doc) Savage via users wrote:
The wired Ethernet never comes up, but as soon as I unplug it the WiFi port's DHCP connection resumes perfectly.
Can anyone suggest the next step?
Are you sure the ethernet network is working? Do you have any other devices using that ethernet? Do you still have a boot entry from F29? Have you tried using it? What are you using for the DHCP server? Can you check if the requests are getting there?
Sam,
Yes, I'm sure the wired Ethernet is working. I have two servers in the other room running fixed IP addresses. The DHCP server is on my U-verse residential gateway, and it's working fine for all WiFi devices. The wired Ethernet was working fine under F29 with the same DHCP service. Also, ethtool verifies that the RJ45 port is working correctly.
The dnf upgrade blew away the old /etc/grub and /boot. I'm wondering if there is some SELinux construct that failed to upgrade completely. As I said, following the F29-F30 upgrade there were a lot of SELinux complaints until I reset the contexts on the / filesystem.
Fedora has evolved so much with NetworkManager, systemd, and other major changes. I've lost track of how a DHCP transaction actually takes place in Fedora now, and what supporting services are at play. Other than journalctl, I don't know where to look any more.
--Doc
On Sat, 2019-05-25 at 23:15 -0700, Samuel Sieb wrote:
On 5/25/19 10:32 PM, Robert G. (Doc) Savage via users wrote:
The wired Ethernet never comes up, but as soon as I unplug it the WiFi port's DHCP connection resumes perfectly.
Can anyone suggest the next step?
Are you sure the ethernet network is working? Do you have any other devices using that ethernet? Do you still have a boot entry from F29? Have you tried using it? What are you using for the DHCP server? Can you check if the requests are getting there? _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
On 5/26/19 5:16 PM, Robert G. (Doc) Savage via users wrote:
Fedora has evolved so much with NetworkManager, systemd, and other major changes. I've lost track of how a DHCP transaction actually takes place in Fedora now, and what supporting services are at play. Other than journalctl, I don't know where to look any more.
Well, in your initial message you said you had the following in your log.
May 25 02:18:33 tiger.protogeek.org NetworkManager[1217]: <info> [1558768713.6863] dhcp4 (enp0s31f6): dhclient started with pid 21697 May 25 02:18:33 tiger.protogeek.org dhclient[21697]: DHCPDISCOVER on enp0s31f6 to 255.255.255.255 port 67 interval 4 (xid=0x94236275) May 25 02:18:37 tiger.protogeek.org dhclient[21697]: DHCPDISCOVER on enp0s31f6 to 255.255.255.255 port 67 interval 4 (xid=0x94236275) May 25 02:18:41 tiger.protogeek.org dhclient[21697]: DHCPDISCOVER on enp0s31f6 to 255.255.255.255 port 67 interval 11 (xid=0x94236275)
A working interface should show something like....
May 26 14:10:57 f30-k.greshko.com NetworkManager[809]: <info> [1558851057.6606] dhcp4 (enp0s3): dhclient started with pid 932 May 26 14:10:58 f30-k.greshko.com dhclient[932]: DHCPDISCOVER on enp0s3 to 255.255.255.255 port 67 interval 6 (xid=0x14de5a6e) May 26 14:11:01 f30-k.greshko.com dhclient[932]: DHCPREQUEST on enp0s3 to 255.255.255.255 port 67 (xid=0x14de5a6e) May 26 14:11:01 f30-k.greshko.com dhclient[932]: DHCPOFFER from 192.168.1.1 May 26 14:11:01 f30-k.greshko.com dhclient[932]: DHCPACK from 192.168.1.1 (xid=0x14de5a6e) May 26 14:11:01 f30-k.greshko.com dhclient[932]: bound to 192.168.1.30 -- renewal in 40477 seconds.
That your system is sending multiple DHCPDISCOVER would seem to indicate that either the system isn't sending anything or the dhcpserver on your LAN isn't receiving it or ignoring it.
I would double check the interface configuration. You may also want to use something like wireshark to verify what is being sent from the interface. Ideally, you could check on the server side as well.
On 5/26/19 1:32 PM, Robert G. (Doc) Savage via users wrote:
Can anyone suggest the next step?
One other thing I would do is....
Normally, a dhcp server is configured to give out only a portion of available IP addresses on a subnet. This leaves you with a set of IP addresses which can be statically assigned.
So, I would reconfigure the interface to use a static address and then make sure it can communicate with other systems on the LAN.
Robert G. (Doc) Savage via users writes:
The dnf upgrade blew away the old /etc/grub and /boot. I'm wondering if there is some SELinux construct that failed to upgrade completely. As I
If you believe that this is SELinux-related, there's a simple way to confirm this. Switch SELinux into permissive mode, and see if this works now.
On 5/26/19 2:16 AM, Robert G. (Doc) Savage via users wrote:
Yes, I'm sure the wired Ethernet is working. I have two servers in the other room running fixed IP addresses. The DHCP server is on my U-verse residential gateway, and it's working fine for all WiFi devices. The wired Ethernet was working fine under F29 with the same DHCP service.
The servers with fixed addresses don't verify that the DHCP is working on the ethernet. Can you use tcpdump on one of them to see if the DHCP broadcasts are visible?
Also, ethtool verifies that the RJ45 port is working correctly.
That just confirms that the driver thinks the port is working, but you already know that because NetworkManager believes the connection is up.
The dnf upgrade blew away the old /etc/grub and /boot. I'm wondering if there is some SELinux construct that failed to upgrade completely. As I said, following the F29-F30 upgrade there were a lot of SELinux complaints until I reset the contexts on the / filesystem.
So you don't have any entries for older kernel versions?
Fedora has evolved so much with NetworkManager, systemd, and other major changes. I've lost track of how a DHCP transaction actually takes place in Fedora now, and what supporting services are at play. Other than journalctl, I don't know where to look any more.
According to the journal and NetworkManager, the DHCP requests are going out, but not getting any response. Have you tried a different ethernet cable? I have had them fail in only one direction before.
-
Ed Greshko -
Robert G. (Doc) Savage -
Sam Varshavchik -
Samuel Sieb