Hello everyone,
I'm trying to determine whether anyone has experienced a similar issue before assuming the hardware is defective.
My setup:
- RB5009UG+S+
- CSS318-16G-2S+ running SwOS
- 10 Gbps DAC connection
- RouterOS and SwOS are up to date
The issue is the following:
When I connect the RB5009 to SFP+1 on the CSS318, the switch starts accumulating:
- Rx MAC Errors
- Rx FCS Errors
Additionally, the link went down once for about 10 seconds and then came back up by itself.
The interesting part is that I moved the same DAC to SFP+2 on the CSS318, and:
- No Rx MAC Errors.
- No Rx FCS Errors.
- The link remains completely stable.
On the RB5009, the interface always reports 10 Gbps Full Duplex, and I don't see any FCS errors on the router side.
I bought this dac https://www.amazon.com/dp/B01LSGGP76?ref=ppx_yo2ov_dt_b_fed_asin_title
My questions are:
- Has anyone seen a case where only one SFP+ port behaves like this?
- Could this still be caused by the DAC even though it works perfectly on SFP+2?
- Is there any known compatibility issue between the RB5009 and the CSS318 (SwOS)?
- Would you suspect a faulty SFP+1 port, or is there anything else I should test before considering an RMA?
Any suggestions or similar experiences would be greatly appreciated.
Thank you!
I don't recall if I have had my CSS318 direct to my RB5009, but I do have a CSS318-16G-2S+, a CSS610-8P-2S+, a CRS305-1G-4S+ and my RB5009UG+S+ all running a 10G backbone with no issues. The CRS305 is on SwOS 2.18 as well, and direct to the RB, and then it fans to the two CSS switches. Early testing had some DAC cables (both 10G and 1.25G) on the CSS318 during migration (the 1.25 to an older NetGear that didn't do 10G), but my interconnects as final are all MultiMode optical with the exception of a short DAC between the CSS610 and the CRS305. The 4th port on my CRS305 has a 1.25G 10GTek RJ45 to and older 1G switch, and other than needing to force the speed no issues there either. Oh, and the RB has been on 7.22/.1/.2 and 7.23.2 - no issues.
Not a solid answer, but I have found no issues in thie mix. 7.23.2 ROS, 2.18 SwOS and (iirc) 2.21 SwOS on the CSS610.
Oh, and the second port on the CSS318 is also in use to an HP DL360G8 with an Intel 560 10G NIC with an (iirc) Finisar optic SFP+, also no issues. (The inter MikroTik links are 10GTek 10G MM optical SFP+ modules.)
Never seen a difference between port behavior on anything. Do you have anything else that you can try on the problematic port?
(The only port errors I see on my CSS318 are some collisions on the two ports running 10MBit to my older UPS gear. I see a few (4 specifically) RX pauses on the CRS305 facing the CSS318)
I'd try a direct test with my spare RB5009 to the CSS318, but I am stuck outnof town at the moment.
Not directly over your target, but hopefully this info helps.
Oh, and nothing but default SFP+ port settings on the RB5009.
Thank you for the information. I just ordered a pair of SFP+ fiber modules to see if the issue still occurs when using fiber. I would really appreciate it if you could perform the test when you have some time. The modules should arrive in about a week. If the problem persists, I will proceed with returning the CSS318 switch.
How much traffic do you need to pass to see errors? As noted, I'll need to use my backup router, since the RB5009 10G port is my backbone, and the two devices are in different rooms. (The CSS318 is in my office, which can come offline briefly to test.)
(I just noted the DAC link, and that is the brand/family I have been using without issue. )
As soon as I connect the DAC cable to the SFP+1 port, the Rx MAC Errors and Rx FCS Errors counters immediately start increasing.
The DAC cable I purchased is this one:
https://www.amazon.com/dp/B01LSGGP76?ref=ppx_yo2ov_dt_b_fed_asin_title&th=1
It is a 10Gtek 10 Gbps passive SFP+ DAC, compatible with Ubiquiti.
Does anything change if you flip it end for end?
And is that "plugged into the CSS and link up, or just the one end plugged in?
(This is starting to sound like a bad SFP port/cage to me, though . . . )
It seems the problem is with the DAC cable itself. I noticed that if I slightly move one end of the cable to the side, the link immediately drops and the connection is lost. When I put the cable back into its normal position, the link comes back and the connection works again.
This behavior appears to indicate that the DAC cable may have an intermittent fault or a bad connection rather than an issue with the switch itself.