Page 1 of 1

CRS Ethernet TX Drops

Posted: Sun Feb 16, 2020 10:44 pm
by tberg
We recently installed a few CRS317 and 328P switches. I am noticing TX Drops incrementing on some Ethernet interfaces. What are TX drops? Are they buffer overflows or filtered traffic with incorrect/unallowed VLAN tags?

Re: CRS Ethernet TX Drops

Posted: Mon Feb 17, 2020 3:57 pm
by tberg
Looking closer, they must be buffer overruns. They are only happening on 1G ports on switches that have 10G ingress and multiple 1G egress. I assume you cannot change interface queue type without breaking hardware offload.

Re: CRS Ethernet TX Drops

Posted: Mon Feb 17, 2020 4:26 pm
by IPANetEngineer
If you are going to be moving a lot of traffic between ports at different speeds , you may be exceeding the buffer (which is required for moving traffic between ports of different speeds.

Re: CRS Ethernet TX Drops

Posted: Mon Feb 17, 2020 6:14 pm
by tberg
Thanks! Yes I'm sure that is it. On the worst port, it is about .1%. We're moving to switch centric architecture and upgrading routers. Once routers are all 10G, part of the problem should go away. However, it's got me thinking about uplink to access switches with 1G backhaul and APs. It might be better to aggregate a few 1G uplinks vs. 10G uplink. Either that or get different switches with bigger buffers?

Re: CRS Ethernet TX Drops

Posted: Thu Apr 02, 2020 6:30 pm
by Maggiore81
Hello
we replaced an UBNT Edgeswitch (all with 1G) ports with a CRS326-24G-2S+ with RouterOS
we just use for VLANS, as a L2 switch.

we replaced the UBNT to take advantage of the S+ ports 10G.
We have now a 10G SFP+ port with DAC to the router (CCR 1009).

We have multiple PTP Wireless licensed uplink coming at 1G speed into the switch, and going aggregated in a VLAN TRUNK port on the CCR through 10G port.
we noticed immediately tx drops.
Is normal or should we change something?