cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
536
Views
1
Helpful
4
Replies

OSPF LSA refresh timer is not 30 minutes

Hello,

I've got very simply topology


R1---R2---R3

RID are respectively

1.1.1.1 2.2.2.2 3.3.3.3


On R1 when I check

show ip ospf topology

I would expect Age timer to reset to zero once it reaches 1800s(Because that's when for example Router2 has to re-send LSA , and timer has to be reset to 0)
But in my case this never happens.
I did a packet capture and was checking Age timer and counter was reset to zero after around 2000 seconds.
Does this mean that at some point Atos changed implementation of LSA Refresh?

1 Accepted Solution

Accepted Solutions

Hello,

Good eye! : ) However, what you observed is normal.

OSPF RFC 2328 does mandate that when an LSA age reaches 1800 seconds, the LSA should be refreshed. However, the same RFC also says that an LSA is valid as long as its age is less than 3600 seconds. So the requirement to refresh an LSA at half its useful life is a very, very conservative approach, and nothing bad happens if the LSA is refreshed somewhat later - as long as the updated LSA gets flooded before the old one expires (due to the age of 3600 and more).

This opens the way for two important approaches: The first is time jittering, meaning that instead of repeating a process at a strict regular interval, you always randomize it a little. This prevents a real phenomenon called global synchronization (don't confuse with TCP global synchronization, though) in which, due to processing delays, the whole network would eventually converge on their timers firing all at the same time.

The second approach is that a router may be the originator of possibly many LSAs that have been originated a few seconds or minutes apart from each other. It makes sense to wait a little and group their refresh together into a single work batch. This is called LSA Group Pacing and you can read about it here:

https://www.cisco.com/c/en/us/td/docs/ios-xml/ios/iproute_ospf/configuration/15-sy/iro-15-sy-book/iro-cfg.html#GUID-CF9E095C-9036-4E7D-82E2-7B471857026D

The bottom line is: What you see is normal on Cisco IOS/IOS XE.

Best regards,
Peter

 

 

View solution in original post

4 Replies 4

Sorry for typo I meant would it be just Cisco who changed timer for LSA Refresh?

Hello,

Good eye! : ) However, what you observed is normal.

OSPF RFC 2328 does mandate that when an LSA age reaches 1800 seconds, the LSA should be refreshed. However, the same RFC also says that an LSA is valid as long as its age is less than 3600 seconds. So the requirement to refresh an LSA at half its useful life is a very, very conservative approach, and nothing bad happens if the LSA is refreshed somewhat later - as long as the updated LSA gets flooded before the old one expires (due to the age of 3600 and more).

This opens the way for two important approaches: The first is time jittering, meaning that instead of repeating a process at a strict regular interval, you always randomize it a little. This prevents a real phenomenon called global synchronization (don't confuse with TCP global synchronization, though) in which, due to processing delays, the whole network would eventually converge on their timers firing all at the same time.

The second approach is that a router may be the originator of possibly many LSAs that have been originated a few seconds or minutes apart from each other. It makes sense to wait a little and group their refresh together into a single work batch. This is called LSA Group Pacing and you can read about it here:

https://www.cisco.com/c/en/us/td/docs/ios-xml/ios/iproute_ospf/configuration/15-sy/iro-15-sy-book/iro-cfg.html#GUID-CF9E095C-9036-4E7D-82E2-7B471857026D

The bottom line is: What you see is normal on Cisco IOS/IOS XE.

Best regards,
Peter

 

 

Thanks a lot for prompt respons

eBardzo dzinkuje za szybka odpowiedz : ) 

Ahoj,

S potešením! : ) (I'm Slovak.)

Všetko dobré praje
Peter

 

Review Cisco Networking for a $25 gift card