Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough? What do I need to do to put the network into the same state as after a fresh reboot?
Many thanks, -T
Maybe, maybe not.
Typically I have had to do a stop(instead of a restart) then a modprobe -r <networkmodulename> ; sleep 1 ; modprobe <networkmodulename> ; then restart.
That works unless the NIC itself has a serious firmware issue and locks up and really needs a hardware reset that comes on a complete reboot.
On Sat, Apr 5, 2025 at 7:04 PM ToddAndMargo via users users@lists.fedoraproject.org wrote:
Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough? What do I need to do to put the network into the same state as after a fresh reboot?
Many thanks, -T
-- _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Sat, Apr 5, 2025 at 7:04 PM ToddAndMargo via users users@lists.fedoraproject.org wrote:
Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough? What do I need to do to put the network into the same state as after a fresh reboot?
Many thanks, -T
On 4/5/25 5:16 PM, Roger Heflin wrote:
Maybe, maybe not.
Typically I have had to do a stop(instead of a restart) then a modprobe -r <networkmodulename> ; sleep 1 ; modprobe <networkmodulename> ; then restart.
That works unless the NIC itself has a serious firmware issue and locks up and really needs a hardware reset that comes on a complete reboot.
Thank you!
ToddAndMargo via users writes:
Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough?
Well? You tell us. Does doing this regain network connectivity, for the problem child in question?
What do I need to do toput the network into the same state as after a fresh reboot?
Well, that really depends on why it's not in a fresh state, to start with. All you know is "it doesn't work". Without knowing the root cause of a busted network connection any advice is just a random, worthless guess.
It could be anything from a broken DHCP server, on the network, refusing to renew the client lease, to a dodgy network card that randomly takes a power, and requires a reboot cycle and a kick in the ass from the kernel, to wake up. It goes without saying that the solution to one would be vastly different than the solution to the other, if a solution is even possible.
On 4/5/25 7:00 PM, Sam Varshavchik wrote:
ToddAndMargo via users writes:
Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough?
Well? You tell us. Does doing this regain network connectivity, for the problem child in question?
What do I need to do to put the network into the same state as after a fresh reboot?
Well, that really depends on why it's not in a fresh state, to start with. All you know is "it doesn't work". Without knowing the root cause of a busted network connection any advice is just a random, worthless guess.
You are under estimating your considerable abilities.
It could be anything from a broken DHCP server, on the network, refusing to renew the client lease, to a dodgy network card that randomly takes a power, and requires a reboot cycle and a kick in the ass from the kernel, to wake up. It goes without saying that the solution to one would be vastly different than the solution to the other, if a solution is even possible.
I removed anydesk-6.3.3.1.el8_x86_64.rpm , which won't work anyway. It thinks I am running Wayland (AnyDesk refuses to support it) and anydesk also throws an "desk_rt_ipc_error".
I am thinking of placing a desktop icon one one of their favorite accounts and having them trigger it, if this happens again:
#!/bin/bash beesu "systemctl stop systemd-networkd.service; sleep 10s; systemctl start systemd-networkd.service"
I am hoping it is not a hardware issue.
Question: Are network cards actually powered off? I am thinking at least part of them has to be on for "Wake on LAN" to operate.
Thank you for the help, -T
On Sat, 2025-04-05 at 20:45 -0700, ToddAndMargo via users wrote:
Question: Are network cards actually powered off? I am thinking at least part of them has to be on for "Wake on LAN" to operate.
That depend on the motherboard (and power supply). When it shuts down it may supply stand-by power to certain things, it may not supply enough current, it may completely shut down, it may be configurable.
On 2025-04-05 19:00, Sam Varshavchik wrote:
ToddAndMargo via users writes:
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough?
Well? You tell us. Does doing this regain network connectivity, for the problem child in question?
What do I need to do to put the network into the same state as after a fresh reboot?
Well, that really depends on why it's not in a fresh state, to start with. All you know is "it doesn't work". Without knowing the root cause of a busted network connection any advice is just a random, worthless guess.
It could be anything from a broken DHCP server, on the network, refusing to renew the client lease, to a dodgy network card that randomly takes a power, and requires a reboot cycle and a kick in the ass from the kernel, to wake up. It goes without saying that the solution to one would be vastly different than the solution to the other, if a solution is even possible.
Exactly this. Have you looked at the logs to see what's happening? Why isn't that the first you do instead of trying to setup hacky workarounds?
On Sat, 2025-04-05 at 17:03 -0700, ToddAndMargo via users wrote:
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
I have a PC that if networking is externally interrupted (A LAN cable is unplugged, the network switch power is interrupted, etc) it does not self-recover. I have to restart that connection on the PC.
Your cause could be external. Someone bumping a cable, loose power socket to a network switch, failing network switch, network switch not liking glitches on the mains, cleaning unplugging something to plug their vacuum into.
If you can't find the cause, perhaps a UPS is in order.
On 4/5/25 9:11 PM, Tim via users wrote:
I have a PC that if networking is externally interrupted (A LAN cable is unplugged, the network switch power is interrupted, etc) it does not self-recover. I have to restart that connection on the PC.
Your cause could be external. Someone bumping a cable, loose power socket to a network switch, failing network switch, network switch not liking glitches on the mains, cleaning unplugging something to plug their vacuum into.
Hmm. The cleaning lady would not be able to get behind the UPS.
But she could be putting inductive spikes on the grid.
If you can't find the cause, perhaps a UPS is in order.
I called out an APC BX1500M. It has been in place all this time.
The hub is getting long the tooth. Maybe I should have that replaced too.
What do you think of this one? https://www.trendnet.com/products/gigabit-switch/24-port-gigabit-desktop-swi...
Tim:
Your cause could be external. Someone bumping a cable, loose power socket to a network switch, failing network switch, network switch not liking glitches on the mains, cleaning unplugging something to plug their vacuum into.
ToddAndMargo:
Hmm. The cleaning lady would not be able to get behind the UPS.
But she could be putting inductive spikes on the grid.
That can happen. It really shouldn't be a problem, though can be, equipment's power supplies should be able to ride out that kind of thing. Older equipment's power supplies may not be as good as they used to be.
If you can't find the cause, perhaps a UPS is in order.
I called out an APC BX1500M. It has been in place all this time.
Is everything on the UPS, computer and hub?
I'd also check all plugs and sockets are firm.
The hub is getting long the tooth. Maybe I should have that replaced too.
What do you think of this one? https://www.trendnet.com/products/gigabit-switch/24-port-gigabit-desktop-swi...
Sorry, unfamiliar with it.
I've had to replace network switches a few times when they spanned between buildings. You could see the LEDs on one of the ports madly blinking when it got damaged (see: "flapping"). Power cycling it would reset things back to normal, but over time failures became permanent.
Outside network cabling is vulnerable to things like somewhat-near lighting, and wind blowing over it causing hight static electricity voltages. And you can have mains wiring problems between buildings, within a building, or even within a room (if it spans a phase) cause problems with equipment joined together by network cabling.
On 4/6/25 3:11 AM, Tim wrote:
Tim:
Your cause could be external. Someone bumping a cable, loose power socket to a network switch, failing network switch, network switch not liking glitches on the mains, cleaning unplugging something to plug their vacuum into.
ToddAndMargo:
Is everything on the UPS, computer and hub?
Server, monitor, hub, firewall, microwave modem. Nothing else.
I'd also check all plugs and sockets are firm.
I was the one that did that. They are all in tight.
The hub is getting long the tooth. Maybe I should have that replaced too.
What do you think of this one? https://www.trendnet.com/products/gigabit-switch/24-port-gigabit-desktop-swi...
Sorry, unfamiliar with it.
I've had to replace network switches a few times when they spanned between buildings. You could see the LEDs on one of the ports madly blinking when it got damaged (see: "flapping"). Power cycling it would reset things back to normal, but over time failures became permanent.
This hub is over 11 yeas old. I never was happy with it to start with.
Outside network cabling is vulnerable to things like somewhat-near lighting, and wind blowing over it causing hight static electricity voltages. And you can have mains wiring problems between buildings, within a building, or even within a room (if it spans a phase) cause problems with equipment joined together by network cabling.
Outside cabling does not affect the internal network. I am able to RDP and AnyDesk into all the workstations when this happens
On 2025-04-06 06:11, Tim via users wrote:
ToddAndMargo:
I called out an APC BX1500M. It has been in place all this time.
Is everything on the UPS, computer and hub?
I'd also check all plugs and sockets are firm.
I once had an APC UPS and a server combination that was causing the server to spontaneously reboot itself. The UPS worked perfectly with any other machine and the server never spontaneously rebooted with any other UPS. Go figure. Probably an oversensitivity of the server's PSU to a non-sinusoidal waveshape of the UPS's output.
Cheers Frank
On Sun, Apr 6, 2025 at 12:37 PM Frank Bures buresf@gmail.com wrote:
On 2025-04-06 06:11, Tim via users wrote:
ToddAndMargo:
I called out an APC BX1500M. It has been in place all this time.
Is everything on the UPS, computer and hub?
I'd also check all plugs and sockets are firm.
I once had an APC UPS and a server combination that was causing the server to spontaneously reboot itself. The UPS worked perfectly with any other machine and the server never spontaneously rebooted with any other UPS. Go figure. Probably an oversensitivity of the server's PSU to a non-sinusoidal waveshape of the UPS's output.
Early APC stuff was very good, but then they went thru a bad patch.
I've seen a few cases where systems running close to the capacity of the UPS would crash.
We were given use of a server with a 750 watt PS that had been used in a number-crunching project. We connected it to 1400 KVA Ferrups UPS, but it would sometimes fail to boot. The numerical code was still installed, and at boot fired up at full speed on all CPU's before it throttled down (thermal constraints?).
Am Sat, 5 Apr 2025 17:03:57 -0700 schrieb ToddAndMargo via users users@lists.fedoraproject.org:
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
Check journalctl --dmesg for any link changes or messages from the NIC's kernel module.
Is it connected to a switch that supports port monitoring? Does the port go down?
On Sat, 2025-04-05 at 17:03 -0700, ToddAndMargo via users wrote:
Problem: about once a week, its networking dies.
I would be looking for a log entry for this. Seeing if you can correlate it to other activity on the machine, or external. Is a regular time? Is it during working hours?
On Sat, Apr 5, 2025 at 9:04 PM ToddAndMargo via users < users@lists.fedoraproject.org> wrote:
Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
That is a workaround -- solution would keep network alive.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Start by checking journalctl for details of the network shutting down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough? What do I need to do to put the network into the same state as after a fresh reboot?
The answer depends on what causes the network to shut down.
I have seen similar issues in the past where the solution came after collecting detailed stats for the times of shutdowns. In one case, the outlet used for a PC was on a circuit used to power an exhaust fan with a timer -- when fan started, voltage dropped to 90V which caused connected gear to reset.
On 4/5/25 5:03 PM, ToddAndMargo via users wrote:
Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough? What do I need to do to put the network into the same state as after a fresh reboot?
Many thanks, -T
Update:
The network froze again.
I had the customer issue a
# systemctl stop systemd-networkd.service # sleep 5s # systemctl start systemd-networkd.service
No joy.
Then I had the customer issue # reboot
No joy.
Then I had him issue a # poweroff and power back on
Joy.
I just ordered out an Intel I210-T1 Network Adapter
-T
On 7 Apr 2025 at 10:59, ToddAndMargo via users wrote:
Date sent: Mon, 7 Apr 2025 10:59:29 -0700 Subject: Re: I need to monitor a dead network To: users@lists.fedoraproject.org Send reply to: Community support for Fedora users users@lists.fedoraproject.org From: ToddAndMargo via users users@lists.fedoraproject.org Copies to: ToddAndMargo ToddAndMargo@zoho.com
On 4/5/25 5:03 PM, ToddAndMargo via users wrote:
Hi All,
Fedora 41
I have a customer I set up a Fedora 41 server.
Problem: about once a week, its networking dies. All network services are dead, xrdp, ssh, samba, etc.. All of them. Console logins work fine.
The solution is to call the customer and have him log in at the console and issue a shutdown. On power up, all work again.
What I would like to do, is to set up a cron job to test the networking (ping firewall) and restart/correct the network if I find it down.
Now I do know about # systemctl restart systemd-networkd.service
But is that enough? What do I need to do to put the network into the same state as after a fresh reboot?
Many thanks, -T
Update:
The network froze again.
I had the customer issue a
# systemctl stop systemd-networkd.service # sleep 5s # systemctl start systemd-networkd.service
No joy.
Then I had the customer issue # reboot
No joy.
Then I had him issue a # poweroff and power back on
Joy.
Just something you might work, but would require changes. I have a number of machines with turbo vnc, but have one machine that sometimes has the port go to closed state. Restarting the service gets it running again.
use nmap to check the vnc port and it if it finds it closed, it restarts the service.
#!/usr/bin/bash x=$(nmap 192.168.10.78 -p 5978 | grep close); if [ -n "$x" ] ; then systemctl restart tvncserver; fi
Could also get info from the network card to see if it has lost the IP address, or if it has gone down.
Could also try a reboot using systemctl reboot Probable be best to find exactly what is going down.
Good Luck.
I just ordered out an Intel I210-T1 Network Adapter
-T
-- _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
+------------------------------------------------------------+ Michael D. Setzer II - Computer Science Instructor (Retired) mailto:mikes@guam.net mailto:msetzerii@gmail.com mailto:msetzerii@gmx.com Guam - Where America's Day Begins G4L Disk Imaging Project maintainer http://sourceforge.net/projects/g4l/ +------------------------------------------------------------+
-
Frank Bures -
George N. White III -
Marco Moock -
Michael D. Setzer II -
Roger Heflin -
Sam Varshavchik -
Samuel Sieb -
Tim -
ToddAndMargo