D-Link Forums
D-Link Enterprise => DGS-1100-Series => Topic started by: J1mbo on August 14, 2019, 11:36:51 AM
-
DGS-1100-05PD (1.00.B033) and DGS-1100-08P (1.00.B031).
BT HG612 PPPoE modem -> 05PD eth1, VLAN 15 untagged
05PD eth4 -> 08P eth1, multiple VLANs trunk ports
08P eth8 -> Unifi EdgeRouter X, multiple VLANs trunk ports
For these 4 ports only, I see steadily increasing RxError counts.
The only commonality if that PPPoE traffic is being carried by all ports. Therefore, I wondered if there could be a bug with the firmware recording RxErrors?
Note: I don't have flow control enabled on these switches.
-
What happens if you remove the ER-X from the mix and try a different router?
I'll try and connect my ER-X and 1100 up and see if I see similar, however I don't have VLANs configured.
-
The RxErrors appear to be a bug relating to jumbo frames. The two switches have jumbo frames enabled and the PPPoE connection is configured with an MTU of 1500 bytes, meaning the MTU between the router and the FTTC modem is 1508.
Running this from a client within the network generates the error counts:
ping -i 0.1 -c 100 -s 1472 -D 8.8.8.8
Conversely, this does not:
ping -i 0.1 -c 100 -s 1460 -D 8.8.8.8
I will raise a ticket with support.
-
Glad you narrowed it down more. Let us know how it turns out.
-
Basically, it's a chipset limitation and the behaviour cannot be fixed. So the port error counters on this switch are essentially useless for any network using jumbo frames. Although the 1100 series are cheap, for me the more correct functionality of the 1200 series would be worth it (along with the cable test functionality).
-
Ya the 1100 is a entry level managed switch so probably not a robust as a mid or high level switch.