I've got some users reporting "499 Try again later - remote server down?" errors when trying to perform a BODY/HEAD/ARTICLE on a newly opened connection. If they then do an XOVER nn-mm on the same connection, these same commands work. Does NNTPCache have come implicit assumption that an XOVER will precede a BODY/HEAD/ARTICLE on a fresh client connection? Is this the error that would be returned if for example the down-stream news server's connection limits had been exceeded? A conventional reader probably normally does this on a single connection. 1) Connect 2) Group 3) XOVER NN-MM 4) Article/Head/Body 5) Article/Head/Body 6) Article/Head/Body 7) Article/Head/Body ... In Newsbin (the program they were using), one connection does 1) Connect 2) Group 3) XOVER NN-MM 4) Disconnect And other connections do 1) Connect 2) Group 3) HEAD/BODY 4) HEAD/BODY 5) HEAD/BODY 6) HEAD/BODY ... 7) Disconnect It sure looks like a sequencing problem.
