OSPF timers

huangedmc · ‎11-27-2009

We'd like to change the default OSPF timers from the default 10/40 to something more aggressive, like 1/4.

All the routing devices are 3560's on a fiber ring, running 1 Gbps, so we figure it's safe to do so.

Thoughts?

Is there a global command to change the timers, or do I have to do it on every interface, even the ones that are put in passive-interface? (ip ospf hello-interval 1)

Also do we need to change it on all the routers/L3 switches in the same area, or can some of them keep the default?

Four of the 3560's will peer w/ CE routers that are managed by our ISP, and they don't want to change the timers...

I saw a similar thread below, but thought I'd start a new one.

https://supportforums.cisco.com/message/723209#723209

Reza Sharifi · ‎11-27-2009

Hi,

The value must be the same for all nodes on your network.

Here is the config guide for 12.2(25) for 3560 switches.

http://www.cisco.com/en/US/docs/switches/lan/catalyst3560/software/release/12.2_25_se/configuration/guide/swiprout.html#wp1039830

http://www.cisco.com/en/US/docs/switches/lan/catalyst3560/software/release/12.2_25_se/configuration/guide/swiprout.html#wp1068572

HTH

Reza

Giuseppe Larosa · ‎11-28-2009

Hello Huang,

>>Is there a global command to change the timers, or do I have to do it on every interface, even the ones that are put in passive-interface? (ip ospf hello-interval 1)

as far as I know there is no global or process level command for setting OSPF timers.

As a result of this you can tune the OSPF timers on a per interface basis.

The reason for the missing global level, process level command is that OSPF uses different default timers on different types of interfaces.

Timers values have to match to build a successful OSPF adjacency.

Recent improvements that may be supported in your switches include sub-second OSPF hello frequencies and BFD for IP fast convergence (the latter is unlikely supported on low end multilayer switches).

Hope to help

Giuseppe

huangedmc · ‎11-29-2009

"The reason for the missing global level, process level command is that OSPF uses different default timers on different types of interfaces.

Timers values have to match to build a successful OSPF adjacency."

My thought was as long as the area ID, md5, netmask, and the timers are the same, two OSPF nodes will form adjacency.

So it's possible that some interfaces have different timers in the same area due to their interface type, or the network engineer purposely changed them correct?

I was a bit consused because Reza said all nodes in a network must have the same timer above...

It looks like BFD command is available on our 3560's when configuring the ospf router process, but when you actually configure the interval on the interface, it gives you a middle finger:

AiPH1-SWHC-2-17(config)#router ospf 1

AiPH1-SWHC-2-17(config-router)#bfd ?
all-interfaces Enable BFD on all interfaces

AiPH1-SWHC-2-17(config-router)#bfd all-interfaces
AiPH1-SWHC-2-17(config-router)#int vlan 800

AiPH1-SWHC-2-17(config-if)#bfd interval 50 min_rx 50 multiplier 3
^
% Invalid input detected at '^' marker.

AiPH1-SWHC-2-17(config-if)#bfd?
% Unrecognized command

What's this sub-second feature you mentioned, and how do I enable it?

Jon Marshall · ‎11-29-2009

huangedmc wrote:

"The reason for the missing global level, process level command is that OSPF uses different default timers on different types of interfaces.

Timers values have to match to build a successful OSPF adjacency."

My thought was as long as the area ID, md5, netmask, and the timers are the same, two OSPF nodes will form adjacency.

So it's possible that some interfaces have different timers in the same area due to their interface type, or the network engineer purposely changed them correct?

I was a bit consused because Reza said all nodes in a network must have the same timer above...

It looks like BFD command is available on our 3560's when configuring the ospf router process, but when you actually configure the interval on the interface, it gives you a middle finger:

AiPH1-SWHC-2-17(config)#router ospf 1

AiPH1-SWHC-2-17(config-router)#bfd ?
all-interfaces Enable BFD on all interfaces

AiPH1-SWHC-2-17(config-router)#bfd all-interfaces
AiPH1-SWHC-2-17(config-router)#int vlan 800

AiPH1-SWHC-2-17(config-if)#bfd interval 50 min_rx 50 multiplier 3
^
% Invalid input detected at '^' marker.

AiPH1-SWHC-2-17(config-if)#bfd?
% Unrecognized command

What's this sub-second feature you mentioned, and how do I enable it?

Have a look at this link for how to configure OSPF fast hellos -

OSPF Fast hellos

Timers have to match between OSPF neigbors per segment to form an adjacency so they are not necessarily the same across the entire network as different network types have different hello/dead interval timers.

Jon

huangedmc · ‎11-29-2009

Thanks; I'll give the fast hello a try.

Does anyone know if I'm supposed to change the timer on ALL the interfaces, or just the non-passive interfaces facing the other routers?

i.e. do I run fast hello even on passive interfaces facing end users?

Also, do Cisco ASA's and Netscreen SSG's support fast hello?

I'm thinking the answer is no, but wanted to confirm...

Nagendra Kumar Nainar · ‎11-29-2009

Hi,

You dont need to change the OSPF timers on passive interface. It needs to be configured only on interface which will involve in neighborship establishment.

FYI, It is better to configure BFD instead of OSPF fast hello as fast hello is high CPU intensive and may trigger fault lalarm at times.

Check if your platform/version supports BFD for OSPF. If yes, go with BFD. You can configure BFD to identify neighborship issue in milliseconds.

HTH,

Nagendra