{"UUID":"c8c96619-9c15-40a7-8087-5b1ecf345adc","URL":"http://www.faqs.org/rfcs/rfc789.html","ArchiveURL":"","Title":"ARPANET network-wide outage due to corrupted routing updates","StartTime":"1980-10-27T00:00:00Z","EndTime":"0001-01-01T00:00:00Z","Categories":null,"Keywords":["arpanet","imp","routing","network control protocols","hardware malfunction","software bug","checksum","sequence numbers"],"Company":"ARPANET","Product":"ARPANET routing protocol","SourcePublishedAt":"0001-01-01T00:00:00Z","SourceFetchedAt":"2026-05-04T17:53:04.757518Z","Summary":"A malfunctioning IMP ([Interface Message Processor](https://en.wikipedia.org/wiki/Interface_Message_Processor)) corrupted routing data, software recomputed checksums propagating bad data with good checksums, incorrect sequence numbers caused buffers to fill, full buffers caused loss of keepalive packets and nodes took themselves off the network. From 1980.","Description":"The ARPANET experienced a network-wide outage on October 27, 1980, lasting several hours. During this period, IMPs (Interface Message Processors) were unable to communicate reliably, leading to \"net trouble\" error messages and broken connections across the network. Attempts to restart individual IMPs were temporarily successful until they reconnected to the network and received the problematic updates.\n\nThe immediate cause was a \"freakish hardware malfunction\" in IMP 50, and possibly IMP 29, which generated faulty network control packets. This led to a high-priority software process running out of control within the IMPs, consuming excessive CPU time and buffer space. IMPs became unable to process essential \"HELLO\" and \"I-HEARD-YOU\" messages, causing lines to be declared down.\n\nThe root cause was identified as a specific routing update (sequence number 44) from IMP 50 being retransmitted with corrupted sequence numbers (40 and 8) due to bit-dropping during recreation from tabled information. The existing checksumming mechanism was insufficient to detect these specific corruptions. These three sequence numbers formed a cyclic relationship where each was considered \"LATER\" than another, causing them to circulate indefinitely and exhaust network resources.\n\nCustomer impact was severe, with the network appearing \"unusable\" nationwide for several hours. Connections were summarily broken, and users received error messages indicating no physical path existed.\n\nInitial remediation involved patching IMPs to disregard updates from IMP 50, which successfully restored network operation. A more permanent fix was proposed to modify the definition of \"LATER\" in the routing algorithm to prevent such cyclic conditions from arising with corrupted sequence numbers. Future hardware improvements were also anticipated to address bit error detection and correction more robustly."}