Wireless Access

dear all,

i have 2 cluster with same configuration WLAN (single SSID for employee),

i've configured each cluster with LMS setting,

lets say cluster "test" with primary lms ip : vrrp ip cluster "test",

and back up lms ip : vrrp ip cluster "sql",

cluster "sql" with primary lms ip : vrrp ip cluster "sql", 

and back up lms ip : vrrp ip cluster "test"

and after i disable port cluster "sql", access point directly failover to controller in cluster "test", but when i enable again port cluster "sql", access point wont move switch ip in controller cluster "sql", even i'm already check the pre-emption lms in each ap group

please advice how to do it, so access point will termination controller in it primary controller 

Hi,

Basiclly your right,

AP Failback/ LMS Preemption

The AP failback feature allows an AP associated with the backup controller (backup LMS) to fail back to the primary controller (primary LMS) if it becomes available.

To configure this feature you must:

• Configure the LMS IP address

• Configure the backup LMS IP address

• Enable LMS preemption

• Configure the LMS hold-down timer ( default 600 seconds)

If configured, the AP monitors the primary controller by sending probes using the PAPI protocol every 60 seconds.  If the AP successfully contacts the primary controller for the entire hold-down period (by default, 600 seconds), it will fail back to the primary controller. If the AP is unsuccessful, the AP maintains its connection to the backup controller, restarts the LMS hold-down timer, and continues monitoring the primary controller.

Can u please share with us the output of the following commnad: ( Of one of the AP's that are in the same AP-Group u testing/ With the same LMS/B.LMS settings.

show ap debug system-status ap-name <name>

dear kdisc98,

here i attached the logs of system status ap

Attachment(s)

show ap debug system-status ap-name.txt   41 KB 1 version

dear kdisc98,

here i attach when the AP's failover to another cluster

Attachment(s)

ap test debug.txt   41 KB 1 version

Recent Control Messages from AP to Controller
---------------------------------------------
Date       Time               Message Description
-------------------------------------------------
Tue Apr 23 14:21:41 2019(178 secs ago): SENT REQ type=STATUS_REPORT len=82 peer=10.17.51.104 seq_num=2 num_attempts=1 rtt=0 secs 040000004D040000000905AC1010010400BEBCF90400000002040000000104000000020607020004000000B406070200040000010407020006070200040000
Tue Apr 23 14:21:29 2019(190 secs ago): SENT REQ type=AP_READY len=25 peer=10.17.51.104 seq_num=1 num_attempts=1 rtt=0 secs 0400000014040000000105AC1010010400BEBCF90400000001
--More-- (q) quit (u) pageup (/) search (n) repeat 
                                                   
Tue Apr 23 14:21:29 2019(190 secs ago): SENT REQ type=HELLO len=1579 peer=10.17.51.104 seq_num=0 num_attempts=1 rtt=0 secs 0400000626040000000005AC1010010400BEBCF90400000000050A113368040000001A040000000004000001660400000001040000016D040000001704000001

It looks like it is trying to find .104

dear cjoseph,

.104 is ip back up lms,

it should try to fallback to primary lms ip which is 172.16.16.110 (vip cluster sql) or 172.16.16.102-103 (ip each MC in cluster sql)

---

@Hudaya wrote:

dear all,

 

i have 2 cluster with same configuration WLAN (single SSID for employee),

 

i've configured each cluster with LMS setting,

 

lets say cluster "test" with primary lms ip : vrrp ip cluster "test",

and back up lms ip : vrrp ip cluster "sql",

 

cluster "sql" with primary lms ip : vrrp ip cluster "sql", 

and back up lms ip : vrrp ip cluster "test"

 

and after i disable port cluster "sql", access point directly failover to controller in cluster "test", but when i enable again port cluster "sql", access point wont move switch ip in controller cluster "sql", even i'm already check the pre-emption lms in each ap group

 

please advice how to do it, so access point will termination controller in it primary controller 

---

That should work.

What version of ArubaOS 8 is this, and what is your holddown timer? EDIT 60 seconds, I see.

On a side note, enabling preemption with two clusters could make it hard to manage what access points should be where.  Controllers in a single cluster are one level of redundancy.  Having a Backup LMS is another level of redundancy.  Having the first cluster preempt the second cluster could be unmanageable in that it would be hard to understand what belongs where.

dear cjoseph,

in this environment running AOS 8.2.2.3 sir,

any advice how should we do about this Ha configuration between this 2 cluster ? because the goal is all cluster can handle each other access point even one of cluster down

Controller Information
----------------------
Item Value
---- -----
Primary LMS 10.17.51.104
Backup LMS 172.16.16.102
HA Standby N/A
Using Primary
LMS Preemption Enabled
Hold-down period 60

How are you disabling the controllers in your test?  Are there only single controllers in your clusters?

Honestly, single controllers rarely fail, much less entire clusters.  If all the controllers in your clusters fail, you probably have bigger issues and your access points might not have connectivity to fail over to a separate cluster.  If they do, all your clients will be forced to get different ip addresses and reauthenticate.  What you do not want is if the path between some of your access points and the initial cluster is degraded, some of your access points randomly fail over to the second cluster, creating a huge issue.

---

@cjoseph wrote:

Controller Information
----------------------
Item Value
---- -----
Primary LMS 10.17.51.104
Backup LMS 172.16.16.102
HA Standby N/A
Using Primary
LMS Preemption Enabled
Hold-down period 60

 

How are you disabling the controllers in your test? 

by unplug port controller to core switch

 

Are there only single controllers in your clusters?

for this test, just for test, we have 1 controller in 1 cluster sir

 

Honestly, single controllers rarely fail, much less entire clusters.  If all the controllers in your clusters fail, you probably have bigger issues and your access points might not have connectivity to fail over to a separate cluster.  If they do, all your clients will be forced to get different ip addresses and reauthenticate.  What you do not want is if the path between some of your access points and the initial cluster is degraded, some of your access points randomly fail over to the second cluster, creating a huge issue.

user concern in power rack or anything sir

---

Primary LMS 10.17.51.104
Backup LMS 172.16.16.102

You have a single controller in a cluster and a single controller NOT in a cluster?

dear cjoseph,

i have these 2 group of cluster for labs sir,

in cluster-3 i've configured for cluster, vrrp, and vip,

so i put in ap group cluster-3 with primary lms ip with vip cluster-3 and backup lms ip with vip cluster sql,

and in ap group cluster sql with primary lms ip vip cluster sql and backup lms ip vip cluster-3