TTL 1 becomes 0 for OSPF and DNS but not dropped why

[lohit prasad]

Hi 

Between two end point minimum TTL value should be set as 2 otherwise neighbour directly connected box will drop the packet when it received packet with TTL value less than 2 i.e. TTL = 1 . 

But for OSPF and DNS, I read that TTL value must be set 1 while sending out of box. Then how come it is processed becuse TTL will become 0 at IP header level and wont go to 89 protocol or UDP/TCP for DNS. then wil not it get dropped ? 

TTL is done at control or data plane level ?

[lohit prasad]

Hi experts 

any reply or help in understanding networking would be helpful ? 

[Rolf Fischer · Accepted solution]

Hi,

a packet with TTL=1 is perfectly valid for the receiving router when it it destined to that receiving router itself:

https://supportforums.cisco.com/discussion/13076746/ttl1-and-router-default-behavior

This cretainly applies to all the link-local control traffic like OSPF.

But for OSPF and DNS, I read that TTL value must be set 1 while sending out of box.

DNS? Where did you read this? This possibly refers to DNS broadcasts, which an layer-3 interface with ip helper configured can convert into unicast. Normally there is no DNS server on the local subnet, so the initial TTL for DNS unicasts has to be much higher than 1.

TTL is done at control or data plane level ?

I would say it depends. An IP packet with TTL<=1 always needs to be processed by the CPU:

http://www.cisco.com/c/en/us/about/security-center/ttl-expiry-attack.html

HTH
Rolf

[lohit prasad]

Thanks. It means if packet belongs to itself then TTL value will be 0 but will be processed  okay.