Date: 31 Dec 1999 01:07:37 -0000 Message-ID: <19991231010737.16203.qmail@cr.yp.to> Mail-Followup-To: namedroppers@internic.net From: "D. J. Bernstein" To: namedroppers@internic.net Subject: Which servers crash on ID reuse? BIND tries to avoid quickly repeating its outgoing query IDs. I've been told that one reason for this is to avoid triggering a crash in some old Apple DNS servers. Does anyone have more information about the buggy DNS implementations? What software has the problem? What version fixed the problem? What are the conditions that produce a crash? What happens if two independent caches (or two people running dig) use the same ID? What happens if BIND sends ID 375 to one of these servers, then sends the next 65000 queries to other servers, then sends ID 375 to the same server again? Are these servers deployed on a noticeable number of machines? Do new server implementors have to worry about this interoperability problem? Is any of this documented in the IETF DNS specifications? ---Dan