Netonix Forums

I found that NTP does not work when unit is loaded with a backup config file, do not recall if this happens from scratch build, after the unit is booted I have to disable NTP and then re-enable it. I also had an issue yesterday with SNMP, the boxes were populated by our tech, I found some typos in them, corrected the typos but our collector was not able to poll data from the unit- after disabling and enabling the SNMP service then it was working properly.

The NTP issue so far has been on multiple switches (primarily the WS-12-250-DC), I have yet to have one successfully work after a bootup- it always requires disable and enable. The backup configs were also used from the same FW version that was on the new switch and it also behaved this way with pre- 1.5.26 FW to a new switch with 1.5.26.

Not a major problem, but if I want to have accurate log files, I find myself needing to check status of NTP each time I am in a switch. This was never an issue PRE- 1.5.26 but I am not certain on which FW release this started.

kiwista wrote:NTP: We started seeing it as of 1.5.25 - however we have primarily gone from 1.5.14 to 1.5.25
I don't see it listed as a bug or fix on 1.5.26 either:
A lot of our upgraded WS series no longer get NTP updates. Initially this was when Mikrotiks were the NTP server, however we have also noticed several on 1.5.25 also not getting NTP updates from x.*countrycode*.pool.ntp.org.
Has anyone else struck this?

I have same issue, disable and enable NTP in the switch and it will work- at least for me. I also just reported this bug

Hi team. Just noting another oddity we've run into. We upgraded 2x WS-8-150-DC switches from 1.5.25 to 1.5.26 for testing and both ran into some sort of webserver problem afterwards as we couldn't reach the webGUI at all.

They were still accessible over SSH and after running the "web_conf_check" tool, we regained access to the webGUI. So maybe the lighttpd.conf file was unhappy after the upgrade. Both switches had logged a segmentation fault directly after a socket assignment error during boot:

Code: Select all

Jan 30 12:10:58 root: upgrading certificate
Jan 1 00:00:00 sysinit: Generating a RSA private key
Jan 1 00:00:01 sysinit: .+++++
Jan 1 00:00:12 sysinit: .......................................+++++
Jan 1 00:00:12 sysinit: writing new private key to '/etc/config/lighttpd.pem'
Jan 1 00:00:12 sysinit: -----
Jan 1 00:00:14 sysinit: 2019-01-01 00:00:14: (network.c.310) can't bind to socket: 0.0.0.0:80 Address already in use
Jan 1 00:00:14 sysinit: Segmentation fault
Jan 1 00:00:16 switch[1513]: Detected warm boot
Jan 1 00:00:19 switch[1510]: temp sensor version 1
Jan 1 00:00:20 root: stopped ntp daemon
Jan 1 00:00:20 root: started ntp daemon
Jan 1 00:00:21 root: sync time via ntp


Neither switch was rebooted before the upgrade, one had about a year of uptime and the other had roughly 8 weeks. Both had plenty of available memory as well we otherwise had no issues.

How do you still have EPRS tab mine is gone?

We are seeing a consistent issue across multiple deployments using Netonix switches (so far observed on 26-port models).
Firmware versions affected:
1.5.23
1.5.24
1.5.25
1.5.26
Scenario:
When the majority (or near majority) of ports are supplying 48V passive PoE to UBNT Wave APs, the switch will enter a soft boot loop.
Behavior:
Switch repeatedly reboots / never fully stabilizes
Appears to be a soft loop (not a hard brick)
Occurs after enabling ports all at once or after reboot under load
Does not occur if you slowly power up ports one at a time but will act up again if switch ever reboots

Workaround:
Downgrading to 1.5.14 immediately resolves the issue
System becomes stable again under the same load conditions
Notes:
This has only been observed so far on 26-port switches
Issue appears tied to high-density 48V PoE load (UBNT Wave APs specifically)
Lower PoE load or mixed device environments do not appear to trigger it
Posting here for visibility in case others are seeing similar behavior.

jimmboy48 wrote:Scenario:
When the majority (or near majority) of ports are supplying 48V passive PoE to UBNT Wave APs, the switch will enter a soft boot loop.
Behavior:
Switch repeatedly reboots / never fully stabilizes
Appears to be a soft loop (not a hard brick)
Occurs after enabling ports all at once or after reboot under load
Does not occur if you slowly power up ports one at a time but will act up again if switch ever reboots

Workaround:
Downgrading to 1.5.14 immediately resolves the issue
System becomes stable again under the same load conditions

Wouldn't using the "Boot Start Delay" on the Power tab be a better workaround than downgrading the firmware?

KBrownConsulting wrote:

jimmboy48 wrote:Scenario:
When the majority (or near majority) of ports are supplying 48V passive PoE to UBNT Wave APs, the switch will enter a soft boot loop.
Behavior:
Switch repeatedly reboots / never fully stabilizes
Appears to be a soft loop (not a hard brick)
Occurs after enabling ports all at once or after reboot under load
Does not occur if you slowly power up ports one at a time but will act up again if switch ever reboots

Workaround:
Downgrading to 1.5.14 immediately resolves the issue
System becomes stable again under the same load conditions

Wouldn't using the "Boot Start Delay" on the Power tab be a better workaround than downgrading the firmware?

I don't want to have to setup boot start delays on 2500 switches. This is a firmware issue since downgrading to 1.5.14 resolves the issue. I would be happy to work with netonix to help replicate this issue so this can get resolved in a future firmware.

jimmboy48 wrote:

KBrownConsulting wrote:

jimmboy48 wrote:Scenario:
When the majority (or near majority) of ports are supplying 48V passive PoE to UBNT Wave APs, the switch will enter a soft boot loop.
Behavior:
Switch repeatedly reboots / never fully stabilizes
Appears to be a soft loop (not a hard brick)
Occurs after enabling ports all at once or after reboot under load
Does not occur if you slowly power up ports one at a time but will act up again if switch ever reboots

Workaround:
Downgrading to 1.5.14 immediately resolves the issue
System becomes stable again under the same load conditions

Wouldn't using the "Boot Start Delay" on the Power tab be a better workaround than downgrading the firmware?

I don't want to have to setup boot start delays on 2500 switches. This is a firmware issue since downgrading to 1.5.14 resolves the issue. I would be happy to work with netonix to help replicate this issue so this can get resolved in a future firmware.

Stephen Is this a known issue at all? Is there is plan to resolve it? I'd be happy to setup a lab and get you guys remote access to the switch if that helps at all. Fixing this issue would really help us out here at Nextlink.

jimmboy48 wrote:We are seeing a consistent issue across multiple deployments using Netonix switches (so far observed on 26-port models). Firmware versions affected: 1.5.23 1.5.24 1.5.25 1.5.26 Scenario: When the majority (or near majority) of ports are supplying 48V passive PoE to UBNT Wave APs, the switch will enter a soft boot loop. Behavior: Switch repeatedly reboots / never fully stabilizes Appears to be a soft loop (not a hard brick) Occurs after enabling ports all at once or after reboot under load Does not occur if you slowly power up ports one at a time but will act up again if switch ever reboots Workaround: Downgrading to 1.5.14 immediately resolves the issue System becomes stable again under the same load conditions Notes: This has only been observed so far on 26-port switches Issue appears tied to high-density 48V PoE load (UBNT Wave APs specifically) Lower PoE load or mixed device environments do not appear to trigger it Posting here for visibility in case others are seeing similar behavior.

How many Wave APs are you putting on the switch?
and how are you powering the switch, direct inline from batts, or via transformer?
It kind of sounds like your hitting the max current capacity, ubiquiti boot do surge higher than once their at idle and can draw up to 24W ea. Also depending on the temperature of your enclosure can reduce the efficiencies of transformers (if above like 45°C). I've seen 12-250-DC powered by 48-24V 5A converters, being perfectly fine in winter and then in summer running into bootloops because the cabinet fans died. Wonder if your experiencing something similar and not a FW issue.

Cdev wrote:

jimmboy48 wrote:We are seeing a consistent issue across multiple deployments using Netonix switches (so far observed on 26-port models). Firmware versions affected: 1.5.23 1.5.24 1.5.25 1.5.26 Scenario: When the majority (or near majority) of ports are supplying 48V passive PoE to UBNT Wave APs, the switch will enter a soft boot loop. Behavior: Switch repeatedly reboots / never fully stabilizes Appears to be a soft loop (not a hard brick) Occurs after enabling ports all at once or after reboot under load Does not occur if you slowly power up ports one at a time but will act up again if switch ever reboots Workaround: Downgrading to 1.5.14 immediately resolves the issue System becomes stable again under the same load conditions Notes: This has only been observed so far on 26-port switches Issue appears tied to high-density 48V PoE load (UBNT Wave APs specifically) Lower PoE load or mixed device environments do not appear to trigger it Posting here for visibility in case others are seeing similar behavior.

How many Wave APs are you putting on the switch?
and how are you powering the switch, direct inline from batts, or via transformer?
It kind of sounds like your hitting the max current capacity, ubiquiti boot do surge higher than once their at idle and can draw up to 24W ea. Also depending on the temperature of your enclosure can reduce the efficiencies of transformers (if above like 45°C). I've seen 12-250-DC powered by 48-24V 5A converters, being perfectly fine in winter and then in summer running into bootloops because the cabinet fans died. Wonder if your experiencing something similar and not a FW issue.

We have 16 Wave APs on that switch. Current steady-state PoE draw is about 202 W out of the 400 W switch budget, with most APs sitting around 9-12 W once booted.
The switch is powered from the 48 V DC plant/load distribution. The switch is showing about 49.7 V on the 48 V board, and the internal 24 V rails are sitting around 25.1-25.3 V.
I agree boot surge is worth considering. If all 16 APs can pull up to ~24 W during startup, that is about 384 W before switch overhead, which is close to the 400 W PoE budget. So steady-state power looks fine, but a simultaneous reboot/PoE startup event could still be tight.
Thermals are warm but the fans are running around 8700-9100 RPM. Current temps are roughly 55 C board, 74 C CPU, 74 C PHY on ports 1-12, and 57 C PHY on ports 13-24.
The switch is currently on 1.5.14 because 1.5.26 was soft rebooting. One thing I’m wondering is whether 1.5.26 changed the boot/PoE enable behavior. For example, if 1.5.26 brings all PoE ports up at once, but 1.5.14 staggers them or has a default delay, that could explain why 1.5.14 is stable while 1.5.26 causes reboots, even with the same AP count and no config change to startup timing.
So I agree power/boot surge is still worth ruling out, but from the current snapshot the switch does not look overloaded in steady state. The difference may be in how the newer firmware handles PoE startup rather than the steady-state load itself.