Wasn't the 500 mile email [1] caused by something similar, i.e. software loading a config file made for a newer version loading it and behaving in an undesirable fashion?
> One of the settings that was set to zero was the timeout to connect to the
remote SMTP server. Some experimentation established that on this
particular machine with its typical load, a zero timeout would abort a
connect call in slightly over three milliseconds.
> One of the settings that was set to zero was the timeout to connect to the remote SMTP server. Some experimentation established that on this particular machine with its typical load, a zero timeout would abort a connect call in slightly over three milliseconds.