cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
781
Views
3
Helpful
9
Replies

Query regarding ARP

mukundh86
Beginner
Beginner

Hello All,

I have a simple query regarding ARP. When the ARP table gets refreshed, do the  active sessions in the switch/router get disconnected and reconnected?

Thanks

Mukundh

3 Accepted Solutions

Accepted Solutions

Peter Paluch
Hall of Fame Cisco Employee Hall of Fame Cisco Employee
Hall of Fame Cisco Employee

Hello Mukundh,

Do you have any specific session in mind?

In any case, during the ARP table refresh, it depends on the implementation of the operating system whether it delays sending a packet until the ARP entry is refreshed, or whether it drops the packet for which the ARP lookup/refresh has been initiated. Thus, from the viewpoint of a single packet, it may get either dropped or delayed. Operating systems may differ in this behavior. You have probably encountered the "first ping lost" symptom when pinging a new IP on a directly connected Ethernet segment - this is caused by IOS temporarily dropping packets.

However, as sessions consist of sequences of packets between devices, a drop of the session depends on whether a reliable or unreliable transport protocol is used, and even with an unreliable protocol, whether a single packet can bring the session down. Telnet, SSH or HTTP sessions are carried over TCP which is specifically designed to correct packet losses. Hence, sessions over TCP should never fail as a result of a loss of a few packets.

Please feel welcome to ask further!

Best regards,

Peter

View solution in original post

sammy_bibs
Beginner
Beginner

As Peter mentioned the upper layers will hold the sessions, whereas the switches between will just default back to their 'hub' behaviour and send received packets out of evey interface (except the received port) until they re-learn their mac-address table (cam) relationships.

The routers in the path will do a similar and hold/drop a packet while it performs an ARP request for its next hop (based on the destination IP of the received packet). It will then forward as normal.

Sam



Sent from Cisco Technical Support Android App

View solution in original post

fb_webuser
Frequent Contributor
Frequent Contributor

When ARP table/Cache is cleared, all the connected devices will create an ARP entry in ARP table. In theory, the should get disconnected for a fraction of seconds but in reality, that loss of connectivity for fraction of seconds is not apparent...

---

Posted by WebUser Neeraj Jagga from Cisco Support Community App

View solution in original post

9 Replies 9

Peter Paluch
Hall of Fame Cisco Employee Hall of Fame Cisco Employee
Hall of Fame Cisco Employee

Hello Mukundh,

Do you have any specific session in mind?

In any case, during the ARP table refresh, it depends on the implementation of the operating system whether it delays sending a packet until the ARP entry is refreshed, or whether it drops the packet for which the ARP lookup/refresh has been initiated. Thus, from the viewpoint of a single packet, it may get either dropped or delayed. Operating systems may differ in this behavior. You have probably encountered the "first ping lost" symptom when pinging a new IP on a directly connected Ethernet segment - this is caused by IOS temporarily dropping packets.

However, as sessions consist of sequences of packets between devices, a drop of the session depends on whether a reliable or unreliable transport protocol is used, and even with an unreliable protocol, whether a single packet can bring the session down. Telnet, SSH or HTTP sessions are carried over TCP which is specifically designed to correct packet losses. Hence, sessions over TCP should never fail as a result of a loss of a few packets.

Please feel welcome to ask further!

Best regards,

Peter

sammy_bibs
Beginner
Beginner

As Peter mentioned the upper layers will hold the sessions, whereas the switches between will just default back to their 'hub' behaviour and send received packets out of evey interface (except the received port) until they re-learn their mac-address table (cam) relationships.

The routers in the path will do a similar and hold/drop a packet while it performs an ARP request for its next hop (based on the destination IP of the received packet). It will then forward as normal.

Sam



Sent from Cisco Technical Support Android App

Peter Paluch
Hall of Fame Cisco Employee Hall of Fame Cisco Employee
Hall of Fame Cisco Employee

Hello Sam,

Thank you for joining! Allow me to have one comment.

the upper layers will hold the sessions, whereas the switches between  will just default back to their 'hub' behaviour and send received  packets out of evey interface (except the received port) until they  re-learn their mac-address table (cam) relationships.

Be careful not to confuse the MAC address tables in switches and ARP tables in IP hosts. They serve different purposes - the MAC address table associates a MAC address with a switchport on which the owner of this MAC is connected, and is used for frame switching. ARP table associates IP addresses and MAC addresses, and is used when encapsulating an IP packet into an Ethernet frame. Either of these tables can be flushed and rebuilt without ever touching the other. These tables get confused quite often, as both deal with MAC addresses prominently, but each of them maps the MAC address to a different object (switchport vs. IP address).

As the original poster asked about dropping sessions during ARP table rebuild, the discussion about switches flooding frames actually answers a different question - what would happen if MAC address table, not the ARP table, was flushed during frame delivery.

Best regards,

Peter

fb_webuser
Frequent Contributor
Frequent Contributor

When ARP table/Cache is cleared, all the connected devices will create an ARP entry in ARP table. In theory, the should get disconnected for a fraction of seconds but in reality, that loss of connectivity for fraction of seconds is not apparent...

---

Posted by WebUser Neeraj Jagga from Cisco Support Community App

sammy_bibs
Beginner
Beginner

Peter'

My bad, just reviewed that post. So to clarify,

The switch (assuming its layer 2) should not be effected as its got no ARP mappings/entries. Just the router or switch if its a MLS will have ARP entries.

But the sessions will remain will they not? The router will just buffer/que the next incoming frame in the session while it sends out and ARP request for the destination L3 devices Mac address? (Maybe one dropped packet (expires during ARP request)).

Sam.

Sent from Cisco Technical Support Android App

Peter Paluch
Hall of Fame Cisco Employee Hall of Fame Cisco Employee
Hall of Fame Cisco Employee

Hello Sam,

Oh, no, do not apologize, and please don't take me as criticizing or nitpicking!

The switch (assuming its layer 2) should not be effected as its got no  ARP mappings/entries. Just the router or switch if its a MLS will have  ARP entries.

Correct, and of course, end hosts (clients, servers, etc.) will also have ARP entries.

But the sessions will remain will they not?

Yes, they should. If using TCP as the transport protocol, the application does not even know that a packet loss (for whatever reason, including a missing ARP entry) occured.

Best regards,

Peter

sammy_bibs
Beginner
Beginner

Peter,

Nitpicking is good, the little details make the bigger picture work, So I thank you. I can see how easily mixed up these two can become as they are quite similar in nature.

I am on these forums to try and learn as much as help, after all everyday is a school day.

Regards, Sam.


Sent from Cisco Technical Support Android App

Peter Paluch
Hall of Fame Cisco Employee Hall of Fame Cisco Employee
Hall of Fame Cisco Employee

Hi Sam,

Thank you! And please feel very welcome here on these forums. It is exactly people like you, able to formulate their opinions and discuss them in such well-informed and good-hearted manner, who make these forums very unique.

Best regards,

Peter

Thank you Sam, Neeraj and Peter for the responses. So it really depends on the transport protocol , TCP or UDP, whether ARP table refreshes cause a disruption.

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Recognize Your Peers