[j-nsp] OSPF question
james list
jameslist72 at gmail.com
Thu Oct 29 05:06:39 EDT 2015
Dear experts,
a customer of mine has the following OSPF scenario:
.........................................................
. .
. area0 area10 totally stub .
. | | .
. | +----+ DR | .
. |---|RT1|--------| .
. | +---+ | .
. | | .
. | +----+ | .
. |----|RT2|------| .
. | +----+ | +------------+ .
. | |----|host5/6/7| .
. | +---+ | +------------+ .
. DR |-----|RT3|-----| .
. | +---+ | .
. | | .
. | +---+ | .
.BDR |------|RT4|----| .
. | +---+ BDR .
. | .
. .
...........................................................
Four MX routers RT1, RT2, RT3 and RT4 running OSPF as ABR (area 0 and area
10 totally stub), on area 10 there are some host talking OSPF (host5/6/7 in
reality are more).
On MX side I see a lot of OSPF retransmissions on area 0 expecially on the
DR-other (RT1 and RT2), I guess it’s due to the following:
1- Host 5 or other send LSA update to all routers on area 10 (RT1, RT2,
RT3, RT4)
2- All routers advertise the update to all the others routers in the same
Ethernet segment
3- If for example RT1 and RT2 simultaneously send the same update towards
RT3, RT3 will send the ACK only the first it gets (let’s say from RT1)
4- So RT2 will wait for ACK, put LSA in the Link State Retransmit List and
after 5 seconds (default LSA retransmit interval), RT2 retransmit to RT3
5- After retransmission RT2 get the ACK from RT3
Am I wrong ?
Does it sounds as reality ?
Should I expect instability, which the customer did not get till now ?
Cheers
James
More information about the juniper-nsp
mailing list