2 ms·
> However, the information was corrupted at the source, and those devices that did understand the attributed detected the corruption and rejected it. No they d
by ExoticPearTree 3y ago
> However, the information was corrupted at the source, and those devices that did understand the attributed detected the corruption and rejected it.
No they did not, they shutdown a whole session down for a malformed announcement. The right decision would have been to ignore the corrupt attribute and maybe not install that particular route.
A logged message saying "Attribute 129 with value 0 is invalid" would have helped operators more than just blindly shutting down sessions through which this was announced.
- fanf2 3y agoThe right decision is specified in RFC 7606, which is usually to treat the malformed announcement as a withdrawal. (BGP attributes cannot in general be discarded safely.)