MikroTik

For some reason, rebooting the switch (via the gui) does not clear this. There are a bunch of MAC addresses listed from computers that have not been connected for months.

What can I do about this?

I agree, sadly it also doesnt solve "INTERNAL ERROR:SyntaxError: missing ] after element list" error so I think it somehow connected.
IMO we have to wait for new SwOS version, but question is: how long?

Rebooting the switch doesn't appear to reboot the switch. The uptime stays the same.

Wasn't that a problem back with 1.4, and an upgrade to 1.5 (you have to manually power cycle after the upload since the reboot button is useless) fixes that?

Wasn't that a problem back with 1.4, and an upgrade to 1.5 (you have to manually power cycle after the upload since the reboot button is useless) fixes that?

I just tried "Reboot" on a switch running 1.5.

When "Reboot" is pressed, there is a few seconds delay, then the page attempts to reload, but finishes loading with "Loading..." at the top of the page. Nothing else.

If I then manually refresh the page, the uptime is reported as 31+ days....

Also, if I then hit the "Hosts" tab, I get: "INTERNAL ERROR:SyntaxError: missing ] after element list"

This is a known, reported, unfixed problem with SWOS. I emailed support with all the details.

There is a long standing problem with SWOS where it (the switch) closes connections before all of the data is transferred. This problem is particularly noticed if the switch is a considerable distance from you and hence there could be significant latency in the data transmission from you to switch and back. The reason for the SyntaxError is that the hosts table is only partially transferred. The js function that parses the data gags on it. So, naturally, a switch that is up for a long period of time could have a larger hosts table and therefore has a higher probability of SyntaxError.

The switch I am testing is located on the East Coast of the US and I am on the West Coast. Ping times are about 120ms. It is unmanageable.

What I do at times is find a PC closer to the switch and use a remote desktop solution to manage the device. I tried that just now.

If I hit "Reboot" from the local machine, the page refreshes properly, but the uptime is still 31+ days. Apparently no reboot. Unless the definition of "uptime" is "Since power was applied". Usually it means the time since last cold or warm start.

The local machine is also unable to retrieve the hosts table and results in a SyntaxError. The local machine is in the same city.

I assume if the switch is power cycled, the hosts page will load as it is very small at that point.

This is not a new problem, it has been reported and remains unfixed.

I have a tech that will be at this site today and I will try to repeat the test with the client on the same LAN.

If that doesn't work we will then power cycle the switch.

I'm curious what you find. I was going to try it on 1.5 but decided against it

My only SwOS device is on my home network on the West coast and I'm currently on the East coast - the wife acceptance factor of remote reboots that might interrupt Netflix streaming until I fly back is fairly low.

@fewi: I know that acceptance factor all too well.....

So the plot thickens.

I had the tech connect locally and he couldn't load the hosts page. SyntaxError.
He was on the same lan, so latency was not an issue.

I then had him reboot using the button on the system page and the page refreshed and the uptime reported 20 seconds. Progress!

However, and this is the big surprise. The hosts tab still gave him and me both a SyntaxError even after reboot.

This seems to imply that the reboot button does not clear the hosts (mac fdb) table. Does that mean the switch chip is not actually reset during reboot? That would be a major oversight.

I then had him power cycle the switch and we could then both access the Hosts tab, albeit it had only a few entries. A few minutes later and now I cannot reliably access the switch nor Hosts tab again remotely while a local PC can read the Hosts tab.

Quite often, engine.js does not load completely remotely.

you could try to reboot it from cron

curl -s -m 1 -D - -u admin:password--digest "192.168.ip.ip/reboot" > /dev/null

...
"rebooting the switch (via the gui) does not clear this"
so that's problem

Since it is only on my home LAN, this bug is not critical but it is quite annoying. But, the thing is dirt cheap so you get what you pay for ofcourse.

I'm actually a bit tempted to see whether this switch truly offers wire speed with VLANs set. I do hope so because if not, that would be very unacceptable.