[Solved] CentOS Internet Problem

[sdato]

I am running CentOS 8.1.1911 and I am experiencing a weird network problem. I can ping the server's default gateway and can ping Internet IP addresses (by name or by IP) but DNF, wget and other operations are timing out. The firewall is disabled in the server.

Code: Select all

$ cat /etc/system-release
CentOS Linux release 8.1.1911 (Core)

Routing table (Private IP addressing so not really care about hiding it.)

Code: Select all

$ route -n
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
0.0.0.0         192.168.52.1    0.0.0.0         UG    425    0        0 inside
192.168.52.0    0.0.0.0         255.255.255.0   U     425    0        0 inside

Configuration of the Bridge interface

Code: Select all

$ nmcli con show inside
connection.id:                          inside
connection.uuid:                        d3be3d2c-0ef5-4bfa-8acb-fdd1f2cc5e51
connection.stable-id:                   --
connection.type:                        bridge
connection.interface-name:              inside
connection.autoconnect:                 yes
.
.
.
ipv4.method:                            manual
ipv4.dns:                               192.168.52.1
ipv4.dns-search:                        --
ipv4.dns-options:                       --
ipv4.dns-priority:                      0
ipv4.addresses:                         192.168.52.6/24
ipv4.gateway:                           192.168.52.1

Ping by DNS name:

Code: Select all

$ ping centos.org
PING centos.org (81.171.33.201) 56(84) bytes of data.
64 bytes from ip-81.171.33.201.centos.org (81.171.33.201): icmp_seq=1 ttl=46 time=205 ms
64 bytes from ip-81.171.33.201.centos.org (81.171.33.201): icmp_seq=2 ttl=46 time=208 ms
64 bytes from ip-81.171.33.201.centos.org (81.171.33.201): icmp_seq=3 ttl=46 time=204 ms
64 bytes from ip-81.171.33.201.centos.org (81.171.33.201): icmp_seq=4 ttl=46 time=199 ms
64 bytes from ip-81.171.33.201.centos.org (81.171.33.201): icmp_seq=5 ttl=46 time=199 ms
^C
--- centos.org ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 144ms
rtt min/avg/max/mdev = 199.041/203.150/208.080/3.544 ms

DNF fails:

Code: Select all

$ sudo dnf update
CentOS-8 - AppStream                                                                    0.0  B/s |   0  B     00:30
Failed to download metadata for repo 'AppStream'
Error: Failed to download metadata for repo 'AppStream'

This server is running KVM and the default gateway is a virtual pfsense firewall connected to the same inside bridge interface, the other firewall interface is directly connected to the Internet and that NIC is physically connected to the pfsense VM (PCI pass through).

The bridge interface inside has multiple Ethernet adapters and I have other devices like wireless AP and PCs connected to this bridge and all of them can access the Internet. Among the other devices I have Fedora, Windows 10, gaming consoles and other clients all working fine. There are also other VMs in the same bridge that can access the Internet just fine (Ubuntu and other CentOS server).

Code: Select all

$ bridge link show
3: eno2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 master inside state forwarding priority 32 cost 100
4: ens4f0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 master inside state forwarding priority 32 cost 100
5: ens4f1: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 master inside state disabled priority 32 cost 100
6: ens4f2: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 master inside state disabled priority 32 cost 100
7: ens4f3: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 master inside state disabled priority 32 cost 100
10: virbr0-nic: <BROADCAST,MULTICAST> mtu 1500 master virbr0 state disabled priority 32 cost 100
11: vnet0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 master inside state forwarding priority 32 cost 100

I will appreciate any help to fix this problem. Internet in this server used to work some weeks ago when I created this setup.

[sdato]

Problem is Solved.

It was a problem with pfsense. It was solved by disabling Hardware Checksum Offloading. It seems a problem related to the virtio driver in FreeBSD. What was really weird is that all the other devices in the network were able to connect to the Internet just fine and only the KVM host was having problems.