v1.3.8 FINAL bug reports and comments

[wtm]

Not sure if this is a bug or just a glitch on the devices?
After upgrading to 1.3.8, figured I would see the "Flow Control" on my Mimosa B5c's, but they are not showing as ON, although the UBNT devices are showing?

One switch is powering them directly, and another is not. Both do not show the B5c's with FC on ?

[sirhc]

Not sure if this is a bug or just a glitch on the devices?
After upgrading to 1.3.8, figured I would see the "Flow Control" on my Mimosa B5c's, but they are not showing as ON, although the UBNT devices are showing?

One switch is powering them directly, and another is not. Both do not show the B5c's with FC on ?

That is not a bug, the MIMOSA radios do NOT have Flow Control "active"

[mike99]

Have used it on a site, seems to work fine.

So fare, the only problem I had was after applying a simple VLAN change (changing the untag port if I remember right), the GUI was detecting a change in the config every X seconds and asking for reloading of the page again and again. Rebooted the switch and it was fine after that.

[sbyrd]

I thought it might be a FW version issue, but the Ping Watchdog randomly stops working. I have seen it on my switches that run 1.22 and my switch that runs 1.38.


I have a WS-12-250AC switch that is connected to XW Rocket Ti APs. Due to a know issue with UBNT sometimes the LAN port on the AP locks up when running 1G. I have the switch then setup to ping the APs and if they miss 3 pings in 120 second intervals (6 minutes) it power cycles the port.

Anyway, I have been noticing that sometimes it will not cycle the port or report a watchdog failure in the switch. If I then go into the Watchdog and Enable/Disable it will start working again and then after a period of time it will stop working.

[sirhc]

OK so you verified this is an issue in v1.3.8 as that will be the first thing Eric will say to me, he only chases bug reports on the current version.

I am moving this post to the proper thread and creating a bug report in our internal software.

[sbyrd]

OK so you verified this is an issue in v1.3.8 as that will be the first thing Eric will say to me, he only chases bug reports on the current version.

I am moving this post to the proper thread and creating a bug report in our internal software.

Yes it just happened today on my WS-12-250AC that is running 1.38. However, the bug seems to have been around a while for me as I have seen it on versions earlier than 1.38 as well.

Here's is an excerpt from the log.

On Dec 5th the watchdog worked and rebooted the AP. Later on Dec 5th and again on Dec 7th I manually bounced the ports as I noticed the APs were down and did not want to wait on the watchdog. On Dec 10th I was asleep and the watchdog did not engage and I had to manually bounce the port.

"

Code: Select all

 Dec 5 15:15:29 pinger: Failure pinging 172.16.3.18, powering off port 8 for 5 seconds
Dec 5 15:15:30 sSMTP[3331]: Sent mail for mikrotik@ilines.net (221 Service closing transmission channel) uid=0 username=***** outbytes=566
Dec 5 20:51:15 UI: Port 8 bounced by 66.xxx.xxx.xxx
Dec 7 16:16:16 UI: Port 8 bounced by 66.xxx.xxx.xxx
Dec 10 05:51:56 UI: Port 9 bounced by 66.xxx.xxx.xxx
Dec 10 05:52:32 UI: Configuration changed by 66.xxx.xxx.xxx
Dec 10 05:52:32 UI: PingWatchdog 5 Enable: changed from 'Enabled' to 'Disabled'
Dec 10 05:52:35 pinger: starting up with 7 ping watchdogs and 0 scheduled port bouncesDec 10 05:52:41 UI: Configuration changed by 66.xxx.xxx.xxx
Dec 10 05:52:41 UI: PingWatchdog 5 Enable: changed from 'Disabled' to 'Enabled'
Dec 10 05:52:44 pinger: starting up with 8 ping watchdogs and 0 scheduled port bounces

[Eric Stern]

I've reviewed the code and tested it and I don't see any way it could go wrong. I'm going to have to suggest that its not a bug and the AP is working enough that the ping check is successful, so the ping watchdog is not being triggered.

I'd recommend that the next time it happens before you manually bounce it, log into the switch and ping the AP from the command line. The actual command the ping watchdog uses is "ping -c 3 -W 1 <ip>", which pings 3 times and times out after 1 second. Unless there is 100% packet loss the script considers the ping successful.

[lligetfa]

Eric, how does the switch deal with NTP timesync issues? Are there any safeguards to stop schedules and/or anything that has time related counters if the system time cannot be trusted? I am grasping at straws but wondered if an unreliable NTP time server could affect this. Any time counting process that could be thrown for a loop when NTP makes an adjustment?

[sirhc]

Eric, how does the switch deal with NTP timesync issues? Are there any safeguards to stop schedules and/or anything that has time related counters if the system time cannot be trusted? I am grasping at straws but wondered if an unreliable NTP time server could affect this. Any time counting process that could be thrown for a loop when NTP makes an adjustment?

"Currently" the only switches that use Time based events are the SMART DC switches WS-8-250-DC and WS-12-250-DC and yes the time based functions will not work if the time is not set.

[sbyrd]

I've reviewed the code and tested it and I don't see any way it could go wrong. I'm going to have to suggest that its not a bug and the AP is working enough that the ping check is successful, so the ping watchdog is not being triggered.

I'd recommend that the next time it happens before you manually bounce it, log into the switch and ping the AP from the command line. The actual command the ping watchdog uses is "ping -c 3 -W 1 <ip>", which pings 3 times and times out after 1 second. Unless there is 100% packet loss the script considers the ping successful.

I will test it, but I am 100% certain it is 100% packet loss as this is a known issue where XW Titanium rockets have the LAN port randomly "lockup" and stop passing data when running 1G.

Also my switch and APs are in different subnets/vlans so if the tower router cannot ping the AP I do not know how the switch would be able to ping it and get enough replies to not trip the watchdog.

Again I will do as you suggest and report back next time it happens.