Ok TX-Drop EVERYWHERE

RouterOS General
1

Im going to first like my original post..

http://forum.mikrotik.com/t/tx-drop-vs-tx-drop-10gig-to-1gig/172618/1

But ummm I ended up just for the sake of it checking basically every CRS I have out in the wild.. and they span across 6.48.. 7.0.. 7.12.1… I mean everywhere lol.. and I have CRS317, CRS310, CRS328, the combo SFP/SFP+ and also the 24port one with all 10SFP (sorry not going to write out all the model numbers.. but you guys get the picture).. and THEY ALL have tx-drop counters up.. But NO FCS.. great power tx/rx levels..

I use the 10gtek SFP’s I guess.. but I mean I have like 9 of these switches in a stadium with PERFECT fiber connections.. so im a little lost here.. is there anything that can be done about these drops.. do they even mean anything lol..

But also seeing these tx-drop on copper ports.

Any insight would be nice.. id love some peace of mind.. no one likes to see “DROP” in any way shape or form when you’re looking at your switch stats.

2

I was once told not to worry about the TX Drops, I will search this reply…

3

Thank you.. I’ve been losing my mind over this ..

4

Find anything ?

5

Guess not.

Did you find anything yourself? It’s been over a year.

I’ve been experiencing TX Drops Myself on my core switch. One of my core switches in particular also seems to become unresponsive daily. I’m concerned it’s related to the tx drops.

MikroTik gave me the following response:

To help mitigate this issue, we recommend enabling PFIFO (Packet First-In, First-Out) queues as a software-based buffering mechanism. This can improve packet queuing efficiency. Please use the following CLI commands to configure the queues:

/queue type set ethernet-default pfifo-limit=1000
/queue interface set [find where queue!=no-queue] queue=ethernet-default

>

I've been too scared to try it myself, because that would start loading up packets on the CPU rather than the switch chip. The CPU on my switch is scarily bad.