ARP address learning might fail after ARP HOLD NHs are getting added and deleted from ARP entries continuously. Traffic will be dropped due to this issue.

Ping direct connected IP address will fail and the error is seen as below. There is no memory exhaustion situation on device.
 

user@device> ping X.X.X.X

PING X.X.X.X (X.X.X.X): 56 data bytes

ping: sendto: Cannot allocate memory

ping: sendto: Cannot allocate memory

ping: sendto: Cannot allocate memory

^C

--- X.X.X.X ping statistics ---

3 packets transmitted, 0 packets received, 100% packet loss

 

This issue might be seen if the following conditions are met:
* On all Junos platform
* ARP HOLD NHs are getting added and deleted continuously

This issue root cause is related to an internal counter that tracks the number of NH HOLD routes created for ARP Resolution. Under rare conditions, this counter becomes miscalculated, leading to the system believing that the NH HOLD routes have exceeded their maximum allowable value, when in reality, this is not the case. Consequently, this issue results in ARP resolution failure and communication failure. This software defect is tracked and solved by PR#1442815