Why does both L2MTU and MAX-L2MTU exist?
Wouldn't it be simpler if L2MTU was scrapped and keep only MAX-L2MTU as L2MTU?

anyone?

L2MTU is what its actually set to, it will never transmit more than that
Max L2MTU is the hardware limit of what its capable of doing. This varies from model to model, some only support ~2000 others support ~11000 etc

Removing it would be a really bad idea, its extremely useful for determining the max possible MTU size end-to-end since you might want to do ~9100 (for say IP packing) but find there's a router in between only capable of 4000 so it'll just break connectivity

On the other hand, I don't understand why every single vendor doesn't just set L2MTU to its max possible value out of the box. L3 MTU is different, applications actually do look at that value and adjust their transmission size based on that. L2MTU I don't believe anything does, it's just there as a maximum possible value for stacking labels etc, but it doesn't report back when something is dropped, it just silently drops anyway. I have never found any harm in simply setting L2MTU to its max possible value everywhere in our network

I don't understand why every single vendor doesn't just set L2MTU to its max possible value out of the box

this is exactly what i'm asking.

why don't they just ship "actual L2MTU" at the max possible and scrap the "actual L2MTU" altogether, leaving just "max L2MTU"? that way, whatever your "max L2MTU" is you understand it's also your "actual L2MTU". what exactly is the point of having both settings?

Your question got me wondering, so I thought I'd go read about it. See https://wiki.mikrotik.com/wiki/Manual:M ... uterBoards.

Your question got me wondering, so I thought I'd go read about it. See https://wiki.mikrotik.com/wiki/Manual:M ... uterBoards.

this does not address my question at all.
that article is mostly talking about MTU.
my question is regarding why there is both L2MTU and MAX L2MTU.

anyone?

anyone?

On a layer 2 network you may have another switch that cannot admit frames that are quite as large as what the others can. In that case you may want to limit all switches to an L2MTU equal to the least capable device you have. If you limit the L2MTU to a setting that the least capable device supports, the entire network will behave in a similar way and it will be entirely predictable. Otherwise, if you have each device set to the highest L2MTU possible, you may have weird problems like a jumbo frame being able to pass through two switches but be silently dropped by a third switch because its maximum MTU is lower. These issues could be difficult to troubleshoot in some cases, so it makes sense to have this setting there.

Thats all well and good in theory, and that is the case for L3 MTU as there is path MTU discovery as well as response mechanisms to indicate the MTU is too large
But unless i'm mistaken, L2 MTU will always silently drop. And i'm also not aware of any protocols that specifically go looking for the L2 MTU size and use it (which could potentially cause an issue)

If i'm wrong please link to show me otherwise, but I havn't yet encountered a situation where having various L2 MTU sizes larger than X spread throughout the network is a problem
Netonix shits me with this, they set it to 1528 by default (including ethernet headers) which is too low to be of use for what its designed as. Yes it can be set higher to 9600 but it's yet another step to take, another policy to implement, another thing that can be accidentally forgotten to set etc
Mikrotik at 1598 by default is reasonable, thats enough for the vast majority of purposes so its forgiveable. The only possible scenario I can think of to use larger than 1598 is with IP packing so it's reasonable to just go ahead and also up the L2MTU as you typically will use IP packing on a link by link basis anyway and it should be vetted as you go (switches, radio's etc)

However I also see no harm if it was just set to max possible in every case. Since there's no difference between having it a bit too low and it silently dropping, vs too high and the next switch/radio/whatever along the link dropping it. It's the exact same net result. But at least if it was set at max out of the box then you know if you only have MikroTik gear on a link, you're guaranteed packet delivery of whatever those devices are capable, not whats been set in the config

With Mikrotik it seems that having possibility to set L2MTU (apart from having max L2 MTU as informational field) is for historical reasons. AFAIK in SwOS you can't set L2MTU any more, it's always set to maximum. Sure, you can set (L3) MTU and is limited by L2MTU ...

anyone?

On a layer 2 network you may have another switch that cannot admit frames that are quite as large as what the others can. In that case you may want to limit all switches to an L2MTU equal to the least capable device you have. If you limit the L2MTU to a setting that the least capable device supports, the entire network will behave in a similar way and it will be entirely predictable. Otherwise, if you have each device set to the highest L2MTU possible, you may have weird problems like a jumbo frame being able to pass through two switches but be silently dropped by a third switch because its maximum MTU is lower. These issues could be difficult to troubleshoot in some cases, so it makes sense to have this setting there.

that's not L2MTU, that's just MTU. packet size is ultimately defined by MTU, not L2MTU.

i'm talking about L2MTU and MAX L2MTU.

With Mikrotik it seems that having possibility to set L2MTU (apart from having max L2 MTU as informational field) is for historical reasons. AFAIK in SwOS you can't set L2MTU any more, it's always set to maximum. Sure, you can set (L3) MTU and is limited by L2MTU ...

that makes sense.

wish we could change to always default L2MTU to MAX L2MTU.