Management problem via SFP ports

[sirhc]

My guess and this is only a guess as I did not really sit down and map out your VLAN config nor do I know your standard network MTU and other information and delve into this but I think it has more to do with the fact that your passing Tagged frames between switches through the SFP ports.

Not sure that you needed to up the MTU to 1718 but it probably needed to be larger than 1518 with your VLAN configuration. Probably 1526 or 1528 would give you a safe overhead.

Remember it does not hurt for the switch MTU to be larger than you need as the only packets the switch will "create" is the packets coming from the switch to your PC for the UI or CLI which are probably not going to even reach 1500 due to the nature of the packet payloads.

Some switches actually increase the MTU behind the scene automatically so many bytes larger than what you set MTU to because of VLANs and other encapsulation protocols such as tzsp, we are considering this.

By default I set my switch MTU to 1528 because I do use a simple VLAN config as seen int THIS post and I run a standard 1500 MTU across my entire WISP network because with the number of wireless retries that will occur JUMBO Frames can have ad adverse affect on capacity.

Also I suggest NOT increasing the TCP Window size up on a WISP wireless network.

We were again discussing today of changing the default MTU on our switches t 1528.

[mike99]

The answer seem to be on this link:
http://www.microhowto.info/tutorials/80 ... #idp144560
Effect on the MTU

The MTU (maximum transmission unit) of a network interface is the size of the largest block of data that can be sent as a single unit. The standard Ethernet MTU is 1500 bytes at the network layer or 1518 bytes at the link layer, the difference being due to the 14-byte header and 4-byte frame check sequence that enclose the payload of an Ethernet frame.

On a VLAN trunk the need for each frame to be tagged adds a further 4 bytes of link-layer framing. This can be accommodated either by increasing the link-layer MTU or by reducing the network-layer MTU:

To use the standard network-layer MTU of 1500 bytes, the equipment must support a link-layer MTU of at least 1522 bytes.
If the link-layer MTU were limited to the standard value of 1518 bytes then the network-layer MTU would need to be reduced to 1496 bytes to compensate.

Devices with explicit VLAN support are supposed to accommodate a link-layer MTU of at least 1522 bytes, but if you are using generic hardware then it may be necessary to accept a lower value. All devices on a given IP subnet must use the same network-layer MTU, so if you intend to deviate from the standard value of 1500 bytes then you will need to configure all affected machines

Similar considerations apply when using jumbo frames. The link layer MTU is then much larger, but so is the potential payload, so allowance must still be made.

[sirhc]

That is an excellent post but I just want to add that setting the MTU higher on the switch than is needed really does not hurt so long as it is the only device that has a larger MTU.

I think I made a mistake making the standard MTU 1518 instead of 1522 or 1528.

Most of the people having a VLAN issue is from an MTU of 1518 instead of 1522. I chose 1528 just for a little breathing room for any other strange protocol someone maybe using such as TZSP or such.