I have a rb112 with atheros running as Access Point. I have four fixed CPE antenna units (smartBridges) on them and all these four are disconnected, reassociated and connected every 4 hours! Exact 4 hours!
I can´t find nowhere any setting that causes this.
I have another identical AP running with 8 clients not having that problem.
Signal strengts are good (-50dbm range). All antenna´s on relative shor distance of 50-100 meters from AP.
The mother board itself is not shutting down. It´s a dual radio routerboard and the other radio (for a backhaul link) doesn´t have that problem with its link.
Actually just found my other AP has the same problem. Except of 1 Client CPE that is online for 3days already! (And the link on the second radio is up for already 7 days)
So why are some Client units disconnected after exact 4 hours?
I increased the ¨Disconnect Timeout from 3 to 6 seconds but that doesn´t help.
I checked all settings but see not even an option do perform regular disconnections of clients (no, not watchdog. The card stays up.)
Atheros card Radio and clients with the problem are all SmartBridges CPE units and one Lobometrics
Hardly any private AP in the air and only two other AP of myself on relative short distances.
All Ap´s in different channels.
Problem exist on both MT AP and time is exact 4 hours for each Client CPE but between CPE´s can be 1 to 5 mins.
If this is interferance its damn well organised. A stopwatch can´t do better.
I think the problem have to be found somewhere else.
Interferance is a irregular problem, mine is very regular.
This is a “feature” of SmartBridge 0.01.17 software release. Apparently in earlier releases they had problems with CPE equipment locking up, although I never noticed it, so instead of fixing the problem properly they introduced a periodic reset, by default this is every 4 hours.
You can configure the reset to be at 1, 2, 3, 4 or 5 hour intervals, but you can’t switch it off.
You can only remove this “feature” by downgrading to 0.01.14, but then you run the risk of the units locking up due to the underlying bug.
I have just started a program of migrating from LocustWorld Mesh to MikroTik. These SmartBridge resets never seemed to be a problem with LocustWorld software.
MikroTik has a connection disconnect parameter which allows it to only deregister a client after a set time. The problem is this seems to have a maximum value of 15s, which is not quite enough for the SmartBridge CPE to reset and re-establish the connection.
As such it doesn’t appear to be a real problem as there only seems to be a short break in service, annoying, but not catastrophic.
Bottom line is that the SmartBridge kit affected is now obsolete, so unlikely to be any further software releases, so if this is a real issue the only answer is to replace the CPE.
I only have 12 customers on SmartBridge CPE, so far only 4 migrated from LocustWorld to MikroTik. It doesn’t seem to be an issue when the MikroTik router is configured as AP only, but if you try to use WDS on the same radio for backhaul things get a bit flakey. This seems more to do with WDS than specifically SmartBridge.
Thnks for your answer. Actually I have been told the same by SB. Actually I asked them for this 0.01.17 software in the early days when I was running a network on SB units only (No MT at all).
I had two or three units then locking up and their upgrade was the solution then.
Just for the sake of hava all readings the same and the use of the same SB monitor I gave all my SB units this new upgrade.
But then I migrated to a MT AP network with a log a regurlarly check to see it my network is fine.
Now all the units came up with the 4-hourly dissconnection and re-connection (all within 1-5 seconds though) and its cloacking up my log.
They all work fine, clients don´t notice anything but I want a clean log.
So I reset all SB´s now back to the previous software and so fare have no ´locked´ units and my logs are much more clean!
So, the issue is solved and as far as I´m concerned this tread can be closed.