Switching

Hello

My customer is using a Juniper switch, I'm faced with an unknown problem. Help me plz 😞

In the following network configuration:

When a wireless client connected to AP # 1 roams to AP # 2, the wireless client's ARP information is loaded simultaneously on eth0 and eth1 of the L3 switch and the wireless client cannot communicate.

Clearing the ARP cache manually can solve the problem by correctly learning ARP information, but I do not understand that ARP information is not updated automatically. I am wondering if there is anything to check on my Juniper switch.

Hi Jaehoon,

Please check the Junos version and Juniper switch model being used.  We might be hitting this known issue with flooding in a VLAN although the MAC address is learnt - https://prsearch.juniper.net/InfoCenter/index?page=prcontent&id=PR1321612

Could you try to upgrade to a fixed version and confirm if things look better? - Junos 15.1X53-D58, 18.1R1 or the latest JTAC recommended release should do - https://kb.juniper.net/InfoCenter/index?page=content&id=KB21476

Hope this helps.

Regards,
-r.

--------------------------------------------------

If this solves your problem, please mark this post as "Accepted Solution."
Kudos are always appreciated :).

Thanks I will try tomorrow  ~

This is definitely something needs to be reviewed on the L3 switch. From your description, doesn't look to be Junos device. Also never seen similar symptom on Junos device.

Let's take a look on L2 switch (seems to be Juniper). The left side SW connected to AP1 should have the MAC address of host time-out (300 seconds default) when the host left AP1. There's no mechanism in L2 for L2 switch to talk to L3 switch about MAC updates. At end of the day, its L3 switch's decision how to handle this "MAC mobility" scenario. It has to time out the entry on AP1 direction eventually, but if that's not happening, it's likely a bug 

Hello,

---

@JAEHOON wrote:

 

When a wireless client connected to AP # 1 roams to AP # 2, the wireless client's ARP information is loaded simultaneously on eth0 and eth1

This is not possible in JUNOS. JUNOS ARP table always has exactly 1 L2-egress-interface associated with ARP entry.

If You actually mean that ARP update from the client-A who roamed from AP-1 to AP-2 does not rewrite ARP entry in L3 switch, then I'd have to ask - did You actually observe client-A sending ARP req/reply ? Note that non-ARP Ethernet frame with src.MAC of client-A does not rewrite ARP entry, it HAS to be ARP from the client-A.

---

@JAEHOON wrote:

 

Clearing the ARP cache manually can solve the problem by correctly learning ARP information, but I do not understand that ARP information is not updated automatically. 

---

It should update automatically as soon as ARP from client-A is received by L3 switch. Please do "monitor traffic interface blah-blah" on L3 switch to confirm client-A sends ARP when it roams behind AP-2.

---

@JAEHOON wrote:

I am wondering if there is anything to check on my Juniper switch.

---

You can try "arp-l2-validate" knob on L3 switch' IRB interface if You don't have it already

https://www.juniper.net/documentation/en_US/junos/topics/reference/configuration-statement/arp-l2-validate-edit-interfaces.html 

Release Information

Statement introduced in Junos OS Release 13.2R4 for EX9200 switches.

Statement introduced in Junos OS Release 15.1R4 for QFX Series switches and EX4300 switches.

The "flooding"  as suggested by other poster, would actually help You so it is highly unlikely flooding is responsible for the root cause.

HTH

Thx
Alex 

When the Wireless client is moved, please check for the association of AP with the Wireless client.

Check if the GARP update is seen on the new port on the switch by monitoring the traffic.

On the Switch, check for "show ethernet-switching table" and "show arp no-resolve hostname <Wireless client IP> before and after the client moved.

If the GARP is not seen, investigate on the Wireless client. If not we need to investigate why the switch is not updating the ARP entry properly.