Wireless Access

Hello fellow Airheads,

Last weekend I had a cluster (2 controllers) failover tested multiple times, to experience the results in case of a controller outage.
AP failover went well, however several clients lost their connection.

What I noticed was the client that lost connection, actually were still connected to the same SSID, but with a 'no internet access' notification.
When manually disconnecting and reconnecting to the network, all started working again.
I had 2 similar clients connected to the same AP, one loosing its connection while the other one did not notice a thing.

I tested with the following setup:

2x 7220's (cluster with VRRP and MM)
AP 335
8.6.0.9 and 8.6.0.18

HP Elitebook x360 1040 G6
W10
Intel(R) Wi-Fi 6 AX200 160MHz
22.70.0.6 and 22.140.0.3

Does anyone conduct failover testing and, if so, what result do you see in your environment?
If someone would like to require more info to get a better view on this, please let me know.

Do your controllers in the cluster show L2-Connected?
Does this happen to all of the clients? Or just specific clients? Note that AOS clustering does both distribute APs over the available controllers, and independently clients, which means that if a controller is switched off it can be that some clients on the same AP fail over while others stay on the same controller.
Clients that are in the 'no internet connection' state, what role are they in on the controler? Do they show up under 'show users' and/or 'show ap association'?

With a cluster of 2 controllers, there are 4 situations for the client if you cut down to a single controller: AP moves controller or is not affected; and the user anchor controller moves or is not affected. With a L2-Connected cluster, each of the 4 possible failover scenarios should not affect client connectivity in a way that you see. It may be good to re-run the test and collect information like 'show tech-support' before and after the failover; and may be best to get your Aruba partner or Aruba support involved to get the right commands to run and do an analysis of the data.

------------------------------
Herman Robers
------------------------
If you have urgent issues, always contact your Aruba partner, distributor, or Aruba TAC Support. Check https://www.arubanetworks.com/support-services/contact-support/ for how to contact Aruba TAC. Any opinions expressed here are solely my own and not necessarily that of Hewlett Packard Enterprise or Aruba Networks.

In case your problem is solved, please invest the time to post a follow-up with the information on how you solved it. Others can benefit from that.
------------------------------

Original Message:
Sent: Jul 26, 2022 06:17 AM
From: Jeroen Hassing
Subject: Cluster failover

Hello fellow Airheads,

Last weekend I had a cluster (2 controllers) failover tested multiple times, to experience the results in case of a controller outage.
AP failover went well, however several clients lost their connection.

What I noticed was the client that lost connection, actually were still connected to the same SSID, but with a 'no internet access' notification.
When manually disconnecting and reconnecting to the network, all started working again.
I had 2 similar clients connected to the same AP, one loosing its connection while the other one did not notice a thing.

I tested with the following setup:

2x 7220's (cluster with VRRP and MM)
AP 335
8.6.0.9 and 8.6.0.18

HP Elitebook x360 1040 G6
W10
Intel(R) Wi-Fi 6 AX200 160MHz
22.70.0.6 and 22.140.0.3

Does anyone conduct failover testing and, if so, what result do you see in your environment?
If someone would like to require more info to get a better view on this, please let me know.

Hello Herman, the controllers have all VLAN's shared, are in the same broadcast domain and are L2-connected.
The issue applies to our HP EliteBook which are of the same hardware and use the same driver version.

I haven't gathered detailed info but will do so during  the next failover test.
Wanted to first check in the forum if there are similar experiences with controller failover before moving to raising ticket.

Thanks for your feedback.
Original Message:
Sent: Jul 26, 2022 07:41 AM
From: Herman Robers
Subject: Cluster failover

Do your controllers in the cluster show L2-Connected?
Does this happen to all of the clients? Or just specific clients? Note that AOS clustering does both distribute APs over the available controllers, and independently clients, which means that if a controller is switched off it can be that some clients on the same AP fail over while others stay on the same controller.
Clients that are in the 'no internet connection' state, what role are they in on the controler? Do they show up under 'show users' and/or 'show ap association'?

With a cluster of 2 controllers, there are 4 situations for the client if you cut down to a single controller: AP moves controller or is not affected; and the user anchor controller moves or is not affected. With a L2-Connected cluster, each of the 4 possible failover scenarios should not affect client connectivity in a way that you see. It may be good to re-run the test and collect information like 'show tech-support' before and after the failover; and may be best to get your Aruba partner or Aruba support involved to get the right commands to run and do an analysis of the data.

------------------------------
Herman Robers
------------------------
If you have urgent issues, always contact your Aruba partner, distributor, or Aruba TAC Support. Check https://www.arubanetworks.com/support-services/contact-support/ for how to contact Aruba TAC. Any opinions expressed here are solely my own and not necessarily that of Hewlett Packard Enterprise or Aruba Networks.

In case your problem is solved, please invest the time to post a follow-up with the information on how you solved it. Others can benefit from that.

Original Message:
Sent: Jul 26, 2022 06:17 AM
From: Jeroen Hassing
Subject: Cluster failover

Hello fellow Airheads,

Last weekend I had a cluster (2 controllers) failover tested multiple times, to experience the results in case of a controller outage.
AP failover went well, however several clients lost their connection.

What I noticed was the client that lost connection, actually were still connected to the same SSID, but with a 'no internet access' notification.
When manually disconnecting and reconnecting to the network, all started working again.
I had 2 similar clients connected to the same AP, one loosing its connection while the other one did not notice a thing.

I tested with the following setup:

2x 7220's (cluster with VRRP and MM)
AP 335
8.6.0.9 and 8.6.0.18

HP Elitebook x360 1040 G6
W10
Intel(R) Wi-Fi 6 AX200 160MHz
22.70.0.6 and 22.140.0.3

Does anyone conduct failover testing and, if so, what result do you see in your environment?
If someone would like to require more info to get a better view on this, please let me know.

Did you ever resolve this? We just rebooted one of the controllers in our cluster and when it came back up, several clients were still connected to the SSID, with an IP, but didn't have a route out to the internet. They couldn't even ping their subnet IP. A wireless disconnect/reconnect or a reboot fixed it.

We're on 8.7.1.11. I'm wondering if this is resolved in 8.10. If anyone knows, let me know.

Original Message:
Sent: Jul 26, 2022 10:21 AM
From: Jer
Subject: Cluster failover

Hello Herman, the controllers have all VLAN's shared, are in the same broadcast domain and are L2-connected.
The issue applies to our HP EliteBook which are of the same hardware and use the same driver version.

I haven't gathered detailed info but will do so during  the next failover test.
Wanted to first check in the forum if there are similar experiences with controller failover before moving to raising ticket.

Thanks for your feedback.
Original Message:
Sent: Jul 26, 2022 07:41 AM
From: Herman Robers
Subject: Cluster failover

Do your controllers in the cluster show L2-Connected?
Does this happen to all of the clients? Or just specific clients? Note that AOS clustering does both distribute APs over the available controllers, and independently clients, which means that if a controller is switched off it can be that some clients on the same AP fail over while others stay on the same controller.
Clients that are in the 'no internet connection' state, what role are they in on the controler? Do they show up under 'show users' and/or 'show ap association'?

With a cluster of 2 controllers, there are 4 situations for the client if you cut down to a single controller: AP moves controller or is not affected; and the user anchor controller moves or is not affected. With a L2-Connected cluster, each of the 4 possible failover scenarios should not affect client connectivity in a way that you see. It may be good to re-run the test and collect information like 'show tech-support' before and after the failover; and may be best to get your Aruba partner or Aruba support involved to get the right commands to run and do an analysis of the data.

------------------------------
Herman Robers
------------------------
If you have urgent issues, always contact your Aruba partner, distributor, or Aruba TAC Support. Check https://www.arubanetworks.com/support-services/contact-support/ for how to contact Aruba TAC. Any opinions expressed here are solely my own and not necessarily that of Hewlett Packard Enterprise or Aruba Networks.

In case your problem is solved, please invest the time to post a follow-up with the information on how you solved it. Others can benefit from that.

Original Message:
Sent: Jul 26, 2022 06:17 AM
From: Jeroen Hassing
Subject: Cluster failover

Hello fellow Airheads,

Last weekend I had a cluster (2 controllers) failover tested multiple times, to experience the results in case of a controller outage.
AP failover went well, however several clients lost their connection.

What I noticed was the client that lost connection, actually were still connected to the same SSID, but with a 'no internet access' notification.
When manually disconnecting and reconnecting to the network, all started working again.
I had 2 similar clients connected to the same AP, one loosing its connection while the other one did not notice a thing.

I tested with the following setup:

2x 7220's (cluster with VRRP and MM)
AP 335
8.6.0.9 and 8.6.0.18

HP Elitebook x360 1040 G6
W10
Intel(R) Wi-Fi 6 AX200 160MHz
22.70.0.6 and 22.140.0.3

Does anyone conduct failover testing and, if so, what result do you see in your environment?
If someone would like to require more info to get a better view on this, please let me know.

Unfortunately we've never had this issue solved but fortunately we never had our controllers rebooted during business hours.
I was using 8.6.x back then so the issue might already have been resolved since quite some new software releases have become available.

I haven't spend time on this anymore since I've been migration the on-prem WLAN to Central and decommissioned all WLAN controllers/gateways.

Original Message:
Sent: Dec 02, 2024 03:34 PM
From: Nathan Kuhl
Subject: Cluster failover

Did you ever resolve this? We just rebooted one of the controllers in our cluster and when it came back up, several clients were still connected to the SSID, with an IP, but didn't have a route out to the internet. They couldn't even ping their subnet IP. A wireless disconnect/reconnect or a reboot fixed it.

We're on 8.7.1.11. I'm wondering if this is resolved in 8.10. If anyone knows, let me know.

Original Message:
Sent: Jul 26, 2022 10:21 AM
From: Jer
Subject: Cluster failover

Hello Herman, the controllers have all VLAN's shared, are in the same broadcast domain and are L2-connected.
The issue applies to our HP EliteBook which are of the same hardware and use the same driver version.

I haven't gathered detailed info but will do so during  the next failover test.
Wanted to first check in the forum if there are similar experiences with controller failover before moving to raising ticket.

Thanks for your feedback.
Original Message:
Sent: Jul 26, 2022 07:41 AM
From: Herman Robers
Subject: Cluster failover

Do your controllers in the cluster show L2-Connected?
Does this happen to all of the clients? Or just specific clients? Note that AOS clustering does both distribute APs over the available controllers, and independently clients, which means that if a controller is switched off it can be that some clients on the same AP fail over while others stay on the same controller.
Clients that are in the 'no internet connection' state, what role are they in on the controler? Do they show up under 'show users' and/or 'show ap association'?

With a cluster of 2 controllers, there are 4 situations for the client if you cut down to a single controller: AP moves controller or is not affected; and the user anchor controller moves or is not affected. With a L2-Connected cluster, each of the 4 possible failover scenarios should not affect client connectivity in a way that you see. It may be good to re-run the test and collect information like 'show tech-support' before and after the failover; and may be best to get your Aruba partner or Aruba support involved to get the right commands to run and do an analysis of the data.

------------------------------
Herman Robers
------------------------
If you have urgent issues, always contact your Aruba partner, distributor, or Aruba TAC Support. Check https://www.arubanetworks.com/support-services/contact-support/ for how to contact Aruba TAC. Any opinions expressed here are solely my own and not necessarily that of Hewlett Packard Enterprise or Aruba Networks.

In case your problem is solved, please invest the time to post a follow-up with the information on how you solved it. Others can benefit from that.

Original Message:
Sent: Jul 26, 2022 06:17 AM
From: Jeroen Hassing
Subject: Cluster failover

Hello fellow Airheads,

Last weekend I had a cluster (2 controllers) failover tested multiple times, to experience the results in case of a controller outage.
AP failover went well, however several clients lost their connection.

What I noticed was the client that lost connection, actually were still connected to the same SSID, but with a 'no internet access' notification.
When manually disconnecting and reconnecting to the network, all started working again.
I had 2 similar clients connected to the same AP, one loosing its connection while the other one did not notice a thing.

I tested with the following setup:

2x 7220's (cluster with VRRP and MM)
AP 335
8.6.0.9 and 8.6.0.18

HP Elitebook x360 1040 G6
W10
Intel(R) Wi-Fi 6 AX200 160MHz
22.70.0.6 and 22.140.0.3

Does anyone conduct failover testing and, if so, what result do you see in your environment?
If someone would like to require more info to get a better view on this, please let me know.