2

I am working with a Juniper router which has the capacity to hold 60M routes in the RIB and 4M routes in the FIB. Should the router crash when more than 4M routes are being sent to FIB, considering no policies have been defined to limit the routes that can be pushed into FIB?

4

Should the routes crash when more than 4M routes is being sent to FIB, considering no policies have been defined to limit the routes FIB can hold?

  1. If your concern is related to hardware limitation, then the first 4 million FIB entries will be installed and anything after that will not. You should also be receiving quite a lot of alarms indicating you are starting to exhaust your FIB memory.

    In cases where you exceed the memory threshold, related processes are going to crash, core dump, and attempt to recover. This concept is not specific to Juniper, any BSD/*nix based system will react this way when a process runs out of memory.

  2. If your concern is related to a licensing restriction, Juniper does not enforce FIB scale licenses in software. FIB entries will be installed until you reach your respective hardware's limitation, at which point the behavior will align with #1.

If you need more detail, feel free to add a comment and I can update my answer.

  • My concern was the hardware limitation. Although pushing 15M routes is unrealistic, I wanted to check how the router behaves. I was surprised to know the router actually crashed and rebooted itself. Also learned that this is a known behavior! What are your thoughts? – Rakesh Nittur Feb 19 '20 at 15:31
  • Can you tell me what you mean by crashed, did it reboot entirely, do a redundancy switchover, or what? Also, what router model and what line card is setting the 4M FIB limit? I can get more specific if I know those items. – Jordan Head Feb 19 '20 at 16:30
  • Crash meaning core files are created the system rebooted itself. I don't have an explicit rule to limit the FIB, it's the hardware limitation. That's the maximum Juniper said they can support. This is PTX 160C router – Rakesh Nittur Feb 20 '20 at 16:40
  • I've updated my answer to address your concern, though I intentionally left it fairly generic for others. The exact reason as to why the PTX rebooted is something we could discuss in chat as it would require more detail. However, in general if you exhaust the PFE/FIB memory, the device wouldn't be able to forward traffic if it stayed online, so the end result basically the same. – Jordan Head Feb 20 '20 at 20:49
  • 1
    Correct, it is NOT a bug. There is a finite amount of memory assigned to the FIB, if you exceed that, you're going to have problems. Ultimately, it's up to the company/engineer to determine scale requirements and select a product based on those. – Jordan Head Feb 20 '20 at 21:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.