Hi Tom, I agree the wording was clearly not ideal but we are speaking of BGP, a protocol open to the world. The PR indicate that there is no known work around, so I can not see how someone could have thought "it will never happen", after the RIPE 63 experiment. Clearly not all backbone felt the pain, so it would look like that most of them upgraded.
To me, the sentence implies "within normal operations". It seems that the issue was caused by a deliberate attack. I know there is similar bug on older version of JunOS for M7i with FlowSpec routes which crash rpd as I triggered it when developing exabgp. It may even still be there :p However the likeliness of this bug affecting networks is as well "exceedingly unlikely to occur" :D Thomas On 8 Nov 2011, at 09:51, Tom Hill wrote: > On Tue, 2011-11-08 at 09:06 +0000, Neil J. McRae wrote: >> This is not a new issue. Juniper issued an advisory for it back in >> August. > > You can imagine why no-one jumped to upgrade, however: > > "MPCs (Modular Port Concentrators) installed in an MX Series router may > crash upon receipt of very specific and unlikely route prefix > install/delete actions, such as a BGP routing update. The set of route > prefix updates is non-deterministic and **exceedingly unlikely to > occur.**" > > (Emphasis mine) > > Tom >> > > >
