09-03-2008 01:57 PM - edited 03-06-2019 01:09 AM
Hello,
Enclosed is rather straightforward diagram of square network design.
R-1 - R-2
|..........|
R-3 - R-4
Note: Please ignore the dots, since blanks/spaces are being removed.
Addressing scheme:
R1/R2: .0/30
R1/R3: .4/30
R2/R4: .8/30
R3/R4: .12/30
OSPF is running on all interfaces with only one line "network ... 0.0.0.3 ...".
Question: If link between R2 and R4 fails, ping from R1 to R4 is going through R2 instead of R3. Also, I have two equal routes on R1 for R2/R4 subnet, no matter what's happening with the R2/R4 link?
Thanks,
Marko
09-03-2008 04:31 PM
If the link fails between R2 and R4, R1 should stop using it, but only after network convergence. If the link goes "down" at the hardware level, convergence is generally rapid. However, if the link remains "up" but neighbor contact is lost, you'll need to wait for OSPF to detect that. How fast depends on the hello timer settings.
09-04-2008 11:18 AM
You can speed up the detection of a link loss with BFD.
09-04-2008 11:34 AM
NB: If supported.
09-04-2008 12:59 PM
Thank you all for replies!
What about the second part of the question and unneeded load sharing (note that it always results with 50% success)?
Marko
09-05-2008 12:03 AM
If the link is actually down, you should only have one route in the routing table. If you don't want it to load share EVER, set max-paths to 1 under the OSPF router config.
If you have a link down, the network is then effectively a line. If you still have two routes you need to look at why. For that a little more info on the details would help - what are the links - LANs? Serial links?
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide