v1.5.5 Bug Reports and Comments

[Stephen]

Can you add it into a zip file? And send it that way?

[banderson]

hmmm.... sometime during the night, the problem seems to have rectified itself ?? I refreshed the browser and logged back in again and now the throughput graph is animated and the log has stopped its complaining ;)

(continued from earlier log)
.
.
.
Nov 12 17:08:06 monitor: restarting vtss_appl

Nov 12 17:08:20 monitor: restarting vtss_appl

Nov 12 17:08:33 monitor: restarting vtss_appl

Nov 12 17:08:47 monitor: restarting vtss_appl

Nov 12 17:09:00 monitor: restarting vtss_appl

Nov 12 17:09:14 monitor: restarting vtss_appl

Nov 12 17:09:28 monitor: restarting vtss_appl


^^ that's the last one -- so about an hour almost exactly (does that mean anything ?)...

[Stephen]

Well I can only guess but it was running for what appeared to be a few week's before this started so that seem's to rule out any periodic network activity so there must be an outside factor. Is this device accessible from a public IP address?

[banderson]

Private address.

Ya .. dunno what the outside factor is. A cold restart was never performed. Just installed the firmware update from 1.4.7 to 1.5.5 (which performs a warm restart it seems).

[robgnet]

4 Switches, 50 miles apart, on different electric grids rebooted with 1.5.5 at within 5 minutes of each other. What would cause this? One was a 12-250 and the other was a 12-400? Also at another tower on 1.5.2 did not reboot, but the other switch in the same rack on 1.5.5 also rebooted.

[amojak]

Had a creeping memory leak problem with 1.5.4 here which results in a load of problems with traffic once it runs out of ram.

A reboot of the switch resolves it again.

We thought 1.5.4 had resolved this from 1.5.2 , is there a service we need to disable that is behind the memory leak?

Does 1.5.5 resolve it as i cannot see a mention in the documentation.

Also could a watchdog option be added that reboots the switch if it is close to running out of ram? Not ideal but far better than finding out after it has been dropping traffic on/off for hours.

The switch is a WS-12-250-DC

Thanks

Bill

[mike99]

For the mem leak, have you try to disable discovery ?

P.S. : I also saw UDP amplification attack through DIS protocol targeting AirOS devices that have a public IP address (in router mode) with Ubnt discovery protocol enabled. I wounder how it would affect Ubnt DP on Netonix switch.

[Stephen]

amojak,
It would be worth updating to 1.5.5 to check. Some of the fixes where very subtle and could easily have resulted in several different issue's. The one's listed are just the one's that where identified to be the result's of the issue's in <1.5.5

If you still see something strange. It might be worth disabling STP to check if that fixes the issue. There have been some report's that indicate it could be responsible for leaking in certain network topology's.

[amojak]

hi,

I took as look at other netonix switches we have running and the ones that do not exhibit this issue had LLDP turned off, this one had it on for some unknown reason.

So have turned it off and upgraded to 1.5.5.

The other odd issue i have with this switch which may be indicative of a fault or possibly a cable issue is when you reboot or upgrade it, port 1 does not come back up properly.

This leaves the power side working but no data link at all. If you change a parameter on the port and apply it then off it goes and negotiates gigabit without issue.

The cable from it is only about 4 meters long and drives a Ubiquiti prismstation AC, as do 4 other ports on the same switch.

I am thinking that for some reason Port 1 is not being set up properly at boot time, perhaps this could be looked into there? This behaviour was seen with 1.5.2 too.

The only reason of course we see it on this switch, apart from the fact it uses port 1, is due to the memory leak issue which means we have to reboot it more often, which shows the port 1 initialisation issue.

Another thing is the PoE priority, this is more a niggle than anything else. Mikoritk kit the priority is higher as the number decreases, aka prio 1 is higher than prio 5 .

To me this makes sense as 1 would be "first priority", however we only realised recently that one of our installers had set up PoE priority without understanding it is reversed on Netonix at one site, the higher the number the higher the priority.

cheers

Bill

[mrmarria]

Release: v1.5.5
maybe saved me a drive through the snow!

Was running v1.5.2

Decided to test the Watchdog - Power Enable feature to re-enable a AF11-FX set to shutdown 1-7 am during a low wind and sun charge week for my batteries.

Just wanted it to fire up if the alternate path went down.

Not sure why but the AF-11FX didn't come back on next morning. I tried toggling the power etc, removed the Watchdog, removed the Power off time and toggled more. Nothing was working. As a last ditch effort before the drive up to the site, looked and saw the firmware upgrade available, thought why not.

After loading v1.5.5 and reboot all was well! AF-11x decided to cooperate.

I haven't tried re- enabling the timed off hours or Watchdog yet.....

Another time, after I check the cables again.