MTU for VLAN parent device

Started by random1104, October 21, 2021, 07:39:25 PM

Previous topic - Next topic
Print
Go Down Pages1
User actions
October 21, 2021, 07:39:25 PM
Hello!,

I'm tagging/untagging on a OPNSense virtual machine. The host is already configured with MTU=9000 which is transfered to the VM NIC too.

I would like to set the MTU for my vtnet0 interface to 1600 bytes to properly pass 1500bytes payload tagged. The only option I can see is the VLAN interface MTU which is already 1500. How can I define the MTU for the parent device?

October 25, 2021, 09:25:22 AM #1 Last Edit: October 25, 2021, 09:29:42 AM by random1104
Well, tested defining an interface for the parent device. Still no joy:

root@vfw02:~ # ifconfig vtnet4
vtnet4: flags=8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> metric 0 mtu 1600
        options=800a8<VLAN_MTU,JUMBO_MTU,VLAN_HWCSUM,LINKSTATE>
        ether 6e:61:51:08:8c:94
        inet6 fe80::6c61:51ff:fe08:8c94%vtnet4 prefixlen 64 scopeid 0x5
        media: Ethernet 10Gbase-T <full-duplex>
        status: active
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>
root@vfw02:~ #
root@vfw02:~ #
root@vfw02:~ # ifconfig vtnet4_vlan91
vtnet4_vlan91: flags=8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> metric 0 mtu 1500
        options=80000<LINKSTATE>
        ether 6e:61:51:08:8c:94
        inet6 fe80::6c61:51ff:fe08:8c94%vtnet4_vlan91 prefixlen 64 scopeid 0x14
        inet 172.16.1.3 netmask 0xffffff00 broadcast 172.16.1.255
        inet 172.16.1.1 netmask 0xffffff00 broadcast 172.16.1.255 vhid 68
        groups: vlan
        carp: BACKUP vhid 68 advbase 1 advskew 100
        vlan: 91 vlanpcp: 0 parent interface: vtnet4
        media: Ethernet 10Gbase-T <full-duplex>
        status: active
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>

root@vfw01:~ # ifconfig vtnet4
vtnet4: flags=8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> metric 0 mtu 1600
        options=800a8<VLAN_MTU,JUMBO_MTU,VLAN_HWCSUM,LINKSTATE>
        ether 72:ee:e4:b0:fa:14
        inet6 fe80::70ee:e4ff:feb0:fa14%vtnet4 prefixlen 64 scopeid 0x5
        media: Ethernet 10Gbase-T <full-duplex>
        status: active
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>


root@vfw01:~ # ifconfig vtnet4_vlan91
vtnet4_vlan91: flags=8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> metric 0 mtu 1500
        options=80000<LINKSTATE>
        ether 72:ee:e4:b0:fa:14
        inet6 fe80::70ee:e4ff:feb0:fa14%vtnet4_vlan91 prefixlen 64 scopeid 0x14
        inet 172.16.1.2 netmask 0xffffff00 broadcast 172.16.1.255
        inet 172.16.1.1 netmask 0xffffff00 broadcast 172.16.1.255 vhid 68
        groups: vlan
        carp: MASTER vhid 68 advbase 1 advskew 0
        vlan: 91 vlanpcp: 0 parent interface: vtnet4
        media: Ethernet 10Gbase-T <full-duplex>
        status: active
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>

root@vfw01:~ # ping -D -c 3 -s 1500 172.16.1.3
PING 172.16.1.3 (172.16.1.3): 1500 data bytes
ping: sendto: Message too long
ping: sendto: Message too long
ping: sendto: Message too long

--- 172.16.1.3 ping statistics ---
3 packets transmitted, 0 packets received, 100.0% packet loss


Both are VMs running on top of proxmox, devices seem correct (OVS bridge related):

root@pve01:~# ip addr|grep 1600
116: vmbr1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1600 qdisc noqueue state UNKNOWN group default qlen 1000
124: tap102i4: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1600 qdisc pfifo_fast master ovs-system state UNKNOWN group default qlen 1000
129: tap101i4: <BROADCAST,MULTICAST,PROMISC,UP,LOWER_UP> mtu 1600 qdisc pfifo_fast master ovs-system state UNKNOWN group default qlen 1000


Any pointers?
October 25, 2021, 09:29:00 PM #2
why do you believe that either 9K or 15oo is not workingg properly
October 25, 2021, 10:04:53 PM #3
For ping you must remove 18 bytes to test an mtu of 1500 so in your test:

ping -D -c 3 -s 1500 172.16.1.3

This is testing an MTU of 1518.

Please use

ping -D -c 3 -s 1472 172.16.1.3

to test an MTU of 1500.

P
October 26, 2021, 07:18:49 PM #4
Tried to send 1500 bytes of payload (ping test), which is what the servers expect to send/pass. Hence, the need to set MTU=1600 for the trunk interfaces.
October 26, 2021, 09:01:44 PM #5
I dont understand your reply. Can you repeat the test as I indicated and post the output?
October 28, 2021, 04:38:16 PM #6
Quote from: allebone on October 25, 2021, 10:04:53 PM
For ping you must remove 18 bytes to test an mtu of 1500 so in your test:

ping -D -c 3 -s 1500 172.16.1.3

This is testing an MTU of 1518.

Please use

ping -D -c 3 -s 1472 172.16.1.3

to test an MTU of 1500.

P

root@vfw01:~ # ping -D -c 3 -s 1500 172.16.1.3
PING 172.16.1.3 (172.16.1.3): 1500 data bytes
ping: sendto: Message too long
ping: sendto: Message too long
ping: sendto: Message too long

--- 172.16.1.3 ping statistics ---
3 packets transmitted, 0 packets received, 100.0% packet loss


root@vfw01:~ # ping -D -c 3 -s 1472 172.16.1.3
PING 172.16.1.3 (172.16.1.3): 1472 data bytes
1480 bytes from 172.16.1.3: icmp_seq=0 ttl=64 time=0.959 ms
1480 bytes from 172.16.1.3: icmp_seq=1 ttl=64 time=0.351 ms
1480 bytes from 172.16.1.3: icmp_seq=2 ttl=64 time=0.319 ms

--- 172.16.1.3 ping statistics ---
3 packets transmitted, 3 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 0.319/0.543/0.959/0.294 ms
October 28, 2021, 04:58:17 PM #7
Quote from: allebone on October 25, 2021, 10:04:53 PM
For ping you must remove 18 bytes to test an mtu of 1500 so in your test:

ping -D -c 3 -s 1500 172.16.1.3

This is testing an MTU of 1518.

Please use

ping -D -c 3 -s 1472 172.16.1.3

to test an MTU of 1500.

P
While I agree with your reasoning, I like to point out that 1500 - 18 = 1482. One of your numbers is wrong and it is probably not the 1500. ;)
Deciso DEC750
People who think they know everything are a great annoyance to those of us who do. (Isaac Asimov)
October 28, 2021, 10:55:51 PM #8
oops sorry you are right it is 28 not 18. 20 = ip header and 8 = ICMP header.

Regards the test then it seems you have confirmed the MTU is 1500 and there is no issue?

"root@vfw02:~ # ifconfig vtnet4_vlan91
vtnet4_vlan91: flags=8943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST> metric 0 mtu 1500
        options=80000<LINKSTATE>
        ether 6e:61:51:08:8c:94
        inet6 fe80::6c61:51ff:fe08:8c94%vtnet4_vlan91 prefixlen 64 scopeid 0x14
        inet 172.16.1.3 netmask 0xffffff00 broadcast 172.16.1.255
        inet 172.16.1.1 netmask 0xffffff00 broadcast 172.16.1.255 vhid 68
        groups: vlan
        carp: BACKUP vhid 68 advbase 1 advskew 100
        vlan: 91 vlanpcp: 0 parent interface: vtnet4
        media: Ethernet 10Gbase-T <full-duplex>
        status: active
        nd6 options=21<PERFORMNUD,AUTO_LINKLOCAL>"

relevant info from above = mtu 1500 and IP 172.16.1.3

Test info:
root@vfw01:~ # ping -D -c 3 -s 1472 172.16.1.3
PING 172.16.1.3 (172.16.1.3): 1472 data bytes
1480 bytes from 172.16.1.3: icmp_seq=0 ttl=64 time=0.959 ms
1480 bytes from 172.16.1.3: icmp_seq=1 ttl=64 time=0.351 ms
1480 bytes from 172.16.1.3: icmp_seq=2 ttl=64 time=0.319 ms

Message indeed is received and replied to.

Can you explain further what you want to do? Are you wanting to change this interface to MTU 1600?
Print
Go Up Pages1
User actions