Inexplicable Packet Errors on CRS305-1G-4S+IN on Fiber Only

Hi All,

I’m wondering if anyone can help me solve a mystery. I have a CRS305-1G-4S+IN running firmware 2.13 (reportedly latest).

My 10gig network devices are:


  • TrueNAS box (ASRock Rack E3C246D4I-2T that sports two Intel X550 10gig copper interfaces - COPPER).
  • My workstation (Windows 11) (10GTek “Intel X520-1” w/ 10GTek Intel-compatible transceiver - FIBER)
  • Test Box (Windows 11) (genuine Intel X520-2 w/ genuine Intel 10G transceiver) borrowed from work - FIBER.

The issue, fundamentally, is that I see “packet errors” only when running fiber into the MikroTik switch. Not enough to affect performance. But, well, there shouldn’t be any. Especially considering that if I run fiber directly between My Workstation and the Test Box, I can transfer terabyte after terabyte and never see a single packet error. I can also use some borrowed copper-interface 10g transceivers and connect My Workstation and the Test Box to the MikroTik switch via copper instead of fiber, and I will see no “packet errors.” So, again, the packet errors only occur when using fiber connections to the switch.

Things I believe I have ruled out:

  • It is not the 10GTek NIC or transceiver because my initial test was to just swap it out for the genuine Intel ones. There was no change (packet errors persisted).
  • It is not the fiber cables themselves since I have tried multiple cables, shortened the run for testing, used a “clicky pen” type cleaner, and the same cables perform perfectly with no packet errors when used to directly connect My workstation and the Text Box together.
  • It’s not the switch’s capacity since the switch handles the same amount of bandwidth when I switch to copper transceivers.
  • It’s not related directly to the TrueNAS box because I saw the same packet errors with it turned off and with My Workstation and the Test Box connected via fiber through the MikroTik switch (rather than directly to each other).

Incidentally, my testing methodology is to transfer about 500GB between hosts while occasionally checking the following in powershell:


Get-NetAdapterStatistics | ft -Property ifAlias,ifDesc,OutboundPacketErrors,ReceivedPacketErrors

I’ll see anywhere from 50 to 500 packet errors that seem to occur at random during the transfer. Sometimes in clusters. Sometimes spread out. With different NICs. With different transceivers. With different fiber. But not if I replace all the fiber with copper. There are no errors on the MikroTik switch itself.

Can anyone make sense of this? I’m at a loss.

Thank you!

Hurin

Correction: It does happen on copper. But much less frequently. I converted my network completely to copper and while I would get a packet error every few GB transferred on fiber, It happens only every few TB on copper.

Upon discovering this, I then connected my workstation to the NAS device directly via cat6a at 10gig, and was able to transfer all night long (about 12TB) without a single packet error. Thus, it appears to be the switch at fault.

Could this be a one-off faulty switch? Would it be worthwhile ordering another?