Hey James, On 17 April 2018 at 11:25, James Bensley <[email protected]> wrote:
> Also you say you have OSPF and LDP up but if you bring up BGP over > this link you may have issues. BGP packs UPDATE messages up to the TCP > MSS (derived from the link MTU). If you are carrying the full table BGP messages are limited to 4096. Having said that, OPs configuration is broken, and should be fixed. MTU can only change inside router, not inside a link or a switch, it's non-recoverable configuration error. Best MTU practice: a) configuration MTU on physical interface as large are reasonable (you can choose some reasonable LCD, if that's large enough) b) on logical interfaces, configure protocol MTU to what ever MTU desired by far-end and supported by transport Some MTU calculations for what is '1500B MTU' L3 MTU == 1500B L2 MTU == 1518B (6 + 6 + 2 + L3 + 4) L1 MTU == 1538B (1 + 7 + L2 + 12) JNPR MTU == 1514 (6 + 6 + 2 + L3) JNPR ICMP == 1547 (L3 - 20 - 8) All of these are same MTU, and people are using them exchangeably without actually knowing if A and B end agree on the MTU. So not rarely, but actually typically A and B ends on large MTU networks have different MTU configured, when non-standard 1500B MTU is used, as A-B parties are unable to communicate to each other what is the correct number that needs to be configured. Mostly this brokenness is not visible to anyone, because routers are not offered large frames, but if they were, they'd be blackholed. It's damn shame we didn't fix this win ARP/ND, as we seem unable to configure MTU correctly manually. -- ++ytti _______________________________________________ juniper-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/juniper-nsp
