So the setup first. I've got MikroTik rb3011 as my gateway with one WAN and other ports in bridge. On the MT bridge I’ve got two switches.

  • one is cisco sg300
  • and for the building nearby there is optical fiber with media converters connecting MT and two chained switches TP link tl-sg1024d.

Cisco is root bridge.

Problem is that, at least once a day the network on those unmanaged switches completely fails. Which means that if you run wireshark on any clients there is no traffic whatsoever from the rest of the network, including L2. Also on MT Rx on particular interface shows 0 bps. All lights on TP link are green indicating all OK.

What we done so far is check for any loops and there is none, at least we could not find it. Also we analysed traffic to check for any abnormality like arp flooding or excessive multicasting and everything looks OK. I've hit the wall and don't know what more to check. Any idea would be greatly appreciated.

Edit: Just for clarification, everything on cisco is working fine.

  • 4
    Not trying to be rude here, but dump the unmanaged switch and get any switch with some abilities of logging, statistics, etc. I don't think there's anything else we can help you with here, and all the time you put into this is much more valuable than a simple managed switch. – Teun Vink Oct 26 '20 at 19:46
  • 1
    After it fails, what do you do to recover? Reboot the switches? Wait until it resolves itself? Something else? – Ron Trunk Oct 26 '20 at 19:49
  • @Teun Vink I agree with you completely it's also what I said to the 3rd party. – D Kuzmanovic Oct 26 '20 at 20:28
  • @Ron Trunk It won’t resolve itself. Sometimes just rebooting switches resolves issue sometimes i have to disconnect all cables and connect one by one. One more thing that I wonder is can there be a faulty client that can cause this? – D Kuzmanovic Oct 26 '20 at 20:28
  • "Cisco is root bridge" If the Cisco switch is unmanaged, you cannot guarantee it is the root bridge. Unfortunately, all your other devices are off-topic here, and we cannot help with an unmanaged device. – Ron Maupin Oct 26 '20 at 20:41

You can rule out any loops - they'd show up as massive broadcast traffic in Wireshark.

The SG300 is a managed switch. Log in, check the logs for any interface down/up events, spanning tree changes etc.

On the unmanaged switches, there's little more than checking the link LEDs. Don't forget the media converters either. During outage, check if you can see (ping by IP address) other nodes on the same switch (indicating a problem upstream).

You should replace the unmanaged switches by managed ones, giving you much more insight into the network. Also, ditch the media converters and use SFP modules. These allow the switch to monitor the fiber link. Media converters hide a link problem as the switches might not see any fibre link loss (depending on the converters and their settings).

Many years back I've had my share of fun with a pair of AT-MC102XL converters where one wouldn't really link after any power loss. Replaced them with 1000BASE-SX SFPs some 15 years ago and removed the problem.

  • Thanks for replay. But no, they are not managed. Only those with E on the end are. Like sg1024dE are managed :-/ . – D Kuzmanovic Oct 26 '20 at 20:20
  • Yes, edited the answer - I was misled by the "stacked" which probably means "chained" really. – Zac67 Oct 26 '20 at 20:21

Not the answer you're looking for? Browse other questions tagged or ask your own question.