I recently purchased and am using two CRS328-24P-4S+RM switches. I had a heck of a time getting expected performance from five different S+RJ10 transceivers. I don't have 10G client devices on either end, but simply testing 1G performance very much depended on what transceiver was in what SFP port on either switch. I mapped all of the combinations and managed to get two SFPs on either switch, that when linked, would consistently provide 950Mbits (all I could push with existing clients). Other combinations saw between 750-850Mbits, which is really awkward for 10G ports. I'm not sure if there was any real negotiation happening between the S+RJ10s, or at least SwOS 2.7 wasn't showing anything except 10G (and link whether a cable was attached or not, also fun), so I assume 10G was active, and seeing >100Mbits at least means it wasn't at 100M.

In any case, while trying all the combinations of SFP ports and transceivers to figure out what was up with the throughput, it became very obvious if either side was using SFP3, no forwarding occurred. Because this happened on the same SFP port on two different devices, I somewhat assumed there was a software setting or configuration situation that was the culprit, or that there may be a legitimate, documented hardware limitation at play. I've found no mention of either thus far. However, I did notice there was one other post that mentioned having issues with SFP4, but only on one switch.

All five S+RJ10s were identified by SwOS 2.7 as available, showed 10G and link as normal, but they simply did not pass traffic. The same configuration was used on every other SFP port when testing throughput. Are my switches physically borked or do I need to shake the fairies out of SFP3?

Cheers,

there was one other post that mentioned having issues with SFP4, but only on one switch.

Correct thats a topic I started, viewtopic.php?f=17&t=134057 and I only have one CRS328-24P-4S+RM switch.

Could you please try to reboot in RouterOS and see if you can forward / connect through SFP3 although it might not be your intention to use RouterOS?

Yeah, I'll schedule some time to break my network again The LAG across two S+RJ10 links is so freaking flaky, but I *think* it is RSTP as traffic is black holed on one switch/port combination and then the other switch/port combination every ~20-30 seconds. So I was planning on trying RouterOS to see if it too experienced such LAG/RSTP issues. Will report back.

The throughput issue, LAG issue and SFP port issue are known issues in SwOS, all of them will be fixed in upcoming SwOS version. Currently you will have to use RouterOS until the new SwOS version is released. The LAG issue will be fixed in RouterOS 6.43rc33 as well as SwOS 2.8

Ohai! Thanks for the official voice on that. I'm glad I'm not reporting anything new. I will say, it'd be nice if SwOS had a nice big "alpha" label on it. I was just a bit surprised at how unstable those three major items are for something I planned to use as a switch. Not being disparaging, I genuinely would have appreciated a warning.

Since it sounds like RouterOS shouldn't be an issue, I'll sit tight and wait for that update.

Updated to SwOS 2.8 on both switches...

[BETTER] SFP3 works now!

[BETTER] A LAG on two 1G ports to a RouterOS device seems to operate as expected. Can unplug cables and traffic continues, but there is very poor indication the LAG is successful on the SwOS side. Only see "partner MAC". RouterOS is setup for LACP (802.3ad).

[WORSE] Attempting a LAG across 2x S+RJ10 transceivers between both CRS328's took out my entire network; I assume loops happened based on how screwed it was. I had RSTP enabled on both ports/both devices. I tried with both devices set to LAG-active and one device LAG-active, the other LAG-passive. No dice. So LAG across copper 10G SFP+ has become less functional. In 2.7 it took a few port disables/enables and perhaps a reboot, but I could get the LAG up. Never did I see loop behavior.

Have u fix this issued ?

I think i got the same problem

Upgraded to v2.9 and I'm happy to report that the LAG across 2x S+RJ10 transceivers between both CRS328's works great! No more RSTP issues, throughput is solid, packet lost is non-existent!

Also, as stated in another post the Mikrotik 10G multi-mode fiber transceiver is working correctly as well, and I put both the 10G copper ports and 10G fiber ports in the same LAG successfully. Honestly I was just curious what would happen, but it's working well. Probably not advised, but interesting to see.

So it seems my saga is complete. All the issues I encountered since purchase have been resolved. Thanks for pulling through Mikrotik!