Maestro ARP table dynamic learning

[Maximiliano]

Hi all,

I'm unsure if this is the right place to ask, or if it's a trivial question.

We have two MHOs with 6 SGMs, and the other day, while reviewing the output of 'asg_arp --verify,' we noticed that for some IP addresses, only one SGM has the corresponding MAC address. For other IP addresses, the MAC exists in all of the SGMs.

Is this behavior due to how Maestro manages traffic, making one SGM active for a specific session? And how is the traffic balanced across all SGMs?
Or is there another explanation for this?

[Lari_Luoma]

This is something we see all the time, but it's not necessary an indication of any problems.

It might be that the original arp table entry was moved to STALE state and that's why Maestro still reports it (INCOMPLETE and FAILED entries have no IP-address).
First possible option is that the ARP table never grew to a size larger than gc_thresh1 value (1024 or something), the incomplete entries were never cleaned out.
Second option what I can think of is that the size of the above mentioned table was reached, but the garbage collector didn't go over the problematic entry. This is normal because the garbage collector process tries to take as little resources as possible and doesn't necessarily go over each removable entry in the system.

These processes happen per SGM, thus you might see temporarily a different number of ARP entries that leads the arp_verifier tool to fail.