NWA90AX evicts all clients and mesh participants every night

WelshProgrammer
WelshProgrammer Posts: 3
First Comment
edited November 2023 in Nebula

Hello, I have an issue with my root AP disconnecting all clients and refusing reconnect for about 20-30 minutes every night around the same time, about 9PM.

I see many entries in the log like for all different types of client on both radios:
Station:
aa:bb:cc:dd:ee:ff disconnected by STA timeout on Channel: 11, SSID:
MyWiFi, 2.4GHz, Signal: -50dBm, Download/Upload:
149851918/77957818 Bytes, reason 107, Interface: wlan-1-4

Station: xx:xx:xx:xx:xx:xx disconnected by STA reconnect on Channel: 11,
SSID: MyWiFi, 2.4GHz, Signal: -56dBm, Download/Upload:
2811498/2653214 Bytes, reason 108, Interface: wlan-1-4

Station: xx:xx:xx:xx:xx:xx disconnected by STA reconnect on Channel: 36,
SSID: MyWiFi, 5GHz, Signal: -47dBm, Download/Upload: 1807005/1325057
Bytes, reason 108, Interface: wlan-2-1

And also for the mesh on 5GHz:

Station: aa:bb:cc:dd:ee:ff disconnected by STA timeout on Channel: 36, SSID: SG-12345678, 5GHz, Signal: -50dBm, Download/Upload: 7665938222/7989694118 Bytes, reason 107, Interface: wds-2-9

The reason is nearly always 107 or 108, this is usually followed by lots of:

station xx:xx blocked by key handshake fail on Channel: 11, SSID: MyWiFi, 2.4GHz, Signal: -67dBm, Download/Upload: 0/0 Bytes, reason 2, Interface: wlan-1-1

What do the error codes mean?
In the 802.11 standard they are not defined and are “custom".

Eventually it corrects itself but the mesh takes about 40 minutes to fully heal and it's extremely annoying.

It happens on both this version and the previous firmware revision.

  • Power mode: Full
  • Smart mesh: Enabled
  • Configuration status: Up to date
  • Firmware availability: Up to date
  • Current version: V6.29(ACCV.1) (Latest)

The 5GHz radio is restricted to channel 36 in Nebula, DCS interval and schedule are not set (so I assume this means disabled) since only one channel is selected.

2.4GHz is set to 3 channel deployment since the APs are have a small overlap in coverage.

Utilization is normally sub 70% for both radios, Smart steering is off and 802.11d is enabled.

Please help me solve this issue, thanks. WP.

All Replies

  • Zyxel_Judy
    Zyxel_Judy Posts: 915  Zyxel Employee
    First Anniversary 10 Comments Friend Collector First Answer

    Hi there,

    We have received the request. We hope check your organization and event log deeply. So, please enable customer’s Zyxel support by going to the left sidebar > Help > Support request > Invite Zyxel support as administrator, save the changes for us to check.

    Judy

  • Done, thanks.

  • Zyxel_Judy
    Zyxel_Judy Posts: 915  Zyxel Employee
    First Anniversary 10 Comments Friend Collector First Answer
    edited March 2023

    Hi there,

    Based on the event log and access point connectivity, it appears that three NWA50AX repeater APs and their clients have maintained stable connections to the NWA90AX root AP for the past week.

    1/ Regarding the issue of the mesh taking approximately 40 minutes to fully heal before February 24th, we have found that the root AP was offline from 2023-02-21 20:50 to 20:52 (for example), as indicated in the logs which showed "NCAS connected: 204 Server is alive" and "Netconf connection is disconnected". This suggests that the network from AP to NCAS was unstable during that time.

    The offline status of the root AP caused three repeater APs to also go offline. Normally, the repeater APs will reconnect to the root AP once it comes back online. However, if the repeater APs cannot find the root AP for more than 30 minutes, they will reboot by themselves.

    We have been investigating the reason why your repeater APs did not reconnect to the root AP when it came back online.

    2/ Most of the clients experiencing disconnection issues are connected to the KeplerLegacy SSID, which only supports 2.4GHz. The 2.4GHz frequency band can be shared with other wireless services, which can cause interference issues, such as those caused by Bluetooth devices and microwave ovens.

    You mentioned that the 2.4GHz frequency band is set to 3-channel deployment, but DCS is disabled, which means that the APs use specific channels: channel 11 for the Garage & Office AP, channel 1 for the Kitchen AP, and channel 6 for the Upstairs AP.

    We recommend:

    • In case the clients can connect to 5GHz, connect to 5GHz -enabled SSID with less legacy devices, cleaner and faster than 2.4GHz .
    • Configuring the Radio's DCS setting as follows to allow the APs to automatically select the best channel for optimal performance. Please help to observe whether the connection issue improves.

    Judy

  • Zyxel_Judy
    Zyxel_Judy Posts: 915  Zyxel Employee
    First Anniversary 10 Comments Friend Collector First Answer

    Hi there,

    After carefully look into the site, we determined that the Mesh problem was caused by the Repeater do the DHCP renew by itself.

    To resolve the problem, we have released a new datecode firmware. Please let us know when it would be convenient for us to schedule the update on your Nebula site.

    Thank you.

    Judy

  • Hi Judy, thanks for the update.

    Can you schedule it for 2 or 3 AM UTC, any day is fine, just let me know.

    Regards, WP.

  • Zyxel_Judy
    Zyxel_Judy Posts: 915  Zyxel Employee
    First Anniversary 10 Comments Friend Collector First Answer
    edited March 2023

    Hi @WelshProgrammer,

    The datecode firmware for your site has been successfully upgraded on March 9th at 02:00 UTC.

    Please help to observe your network and tell us if there is any update.

    Thank you!

    Judy

Nebula Tips & Tricks