Fedora 43 Server. mDNS works for about 60 seconds, then quits. Also it can't find any local hosts

Ask Fedora
, ,
1

I just spun up a new Fedora 43 Server and I can’t seem to get any of my local machines to recognize it by its hostname – except for the first 60 seconds or so after I restart systemd-resolved. Works great. Then quits. The status of systemd-resolved does NOT change.

This is my /etc/systemd/resolved.conf file:
[Resolve]
LLMNR=no
MulticastDNS=yes

Why do I have LLMNR disabled? I read somewhere that if you have Apple devices (I do) you don’t want to enable it. However, even with it enabled, the same problem exists, so that’s not the issue.

Not sure if this is related but the new server cannot ping any of the local hosts, no matter what. If I type ping mydesktopmachine.local I always get ping: mydesktopmachine.local: Temporary failure in name resolution But on a different computer I can type that command and get a return, no problem. What’s up with that “temporary” nonsense, too? Like, it expects that to change soon?

So… Anyone got any ideas?

EDIT: I did follow the instructions here, BTW. Correct way to enable mDNS on Fedora Server 34? It didn’t work for me.

2

mDNS also needs to be enabled in NetworkManager. Run resolvectl if mDNS is enabled for your network configuration. For example like this

Global
         Protocols: -LLMNR +mDNS -DNSOverTLS DNSSEC=no/unsupported
  resolv.conf mode: stub

Link 2 (eth0)
    Current Scopes: DNS mDNS/IPv4 mDNS/IPv6
         Protocols: +DefaultRoute -LLMNR +mDNS -DNSOverTLS DNSSEC=no/unsupported
Current DNS Server: 84.116.46.21
       DNS Servers: 84.116.46.21 84.116.46.20 2001:730:3e42:1000::53
                    2001:b88:1002::10 2001:b88:1202::10
     Default Route: yes

LLMNR is only useful on a network where LLMNR is actually being used. LLMNR is an alternative to mDNS but should be able to co-exist.

3

This is the output when I type sudo resolvectl

Global
Protocols: -LLMNR +mDNS -DNSOverTLS DNSSEC=no/unsupported
resolv.conf mode: stub

Link 2 (enp2s0)
Current Scopes: DNS mDNS/IPv4 mDNS/IPv6
Protocols: +DefaultRoute -LLMNR +mDNS -DNSOverTLS DNSSEC=no/unsupported
Current DNS Server: 1.0.0.1
DNS Servers: 1.0.0.1 192.168.1.1
Default Route: yes

So, it seems to be enabled there… What else should I check?

4

The firewall setting.

For example, to enable mdns for the zone FedoraWorkstatio run this command

firewall-cmd --permanent --add-service=mdns --zone=FedoraWorkstatio

You can also add logging using this command

firewall-cmd --set-log-denied=all

Run journalctl -k to list the log entries, which are mixed in with the other kernel log lines.

Also, the avahi-daemon.service and avahi-daemon.socket may interfere with the mdns function of systemd-resolved, so it is best to disable those systemd units.

5

Ok, how do I undo that –set-log-denied command?! It’s been like 3 minutes and there are literally thousands of lines! All exactly the same:

`[date/time] hostname kernel: filter_IN_FedoraServer_REJECT: IN=enp2s0 OUT MAC=[big long MAC address]`

I need to turn that off!

6

Never mind – figured it out. firewall-cmd --set-log-denied=off

Anyway – I tried disabling the avahi-daemon and avahi-daemon.socket, then I restarted the systemd-resolved service, and yet again, mDNS lasted about 60 seconds before failing.

This is becoming quite annoying. Why can’t Fedora just have this working out of the box like literally EVERY OTHER DISTRO EVER?!

7

Oh – I skipped right over this earlier…

I have Fedora Server installed – should the zone be FedoraServer instead of FedoraWorkstation?

Thanks.

8

Never mind… I did the firewall-cmd --permanent --add-service=mdns --zone=FedoraServer thing and then restarted the systemd-resolved service. Same thing as before – 60 seconds of mDNS and then nothing. :roll_eyes:

9

Oh wait – I didn’t restart the firewalld service after adding mDNS to it. That did the trick.
whew
Thanks!

10

I think it might help if you report, what processes are listening on mdns port, like:

sudo ss -lup sport = mdns

Please share configuration of authselect:

authselect current
grep hosts: /etc/nsswitch.conf

Use firewall-cmd to list active zones only: firewall-cmd --get-active-zones.

Few examples might help you:

firewall-cmd --query-service=ssh --zone=FedoraServer
firewall-cmd --query-service=mdns --zone=FedoraServer
firewall-cmd --list-services --zone=FedoraServer

Try also commands specific to avahi or resolved to ensure some of them can resolve the name:

resolvectl query $HOSTNAME.local
avahi-resolve -n $HOSTNAME.local
getent ahosts -s resolve $HOSTNAME.local
getent ahosts -s mdns $HOSTNAME.local

Some of above commands might help to find, what works and what does not. However, it might be even hardware problem of your access point or your router. Are you trying it exclusively on wired ethernet or is wireless connection also used? If yes, do you know its vendor and type?

11

oh, so you have got it working and stable in the end?

I prefer especially for remote machines trying the change first in runtime, then verify it has changed and you did not cut yourself out. Then save the change.

sudo firewall-cmd --add-service=mdns --zone=FedoraServer
# firewall change is applied, but not yet permanent
# now verify it has changed positively and you connection still works.
# if yes, continue. If not, ask somebody to reboot you machine :)
sudo firewall-cmd --runtime-to-permanent
# now it will become active even after the reboot.

This is much safer way to reconfigure network services, where you access it over the network itself. On local devices you can use local console.