On 12 Dec 98, at 11:00, Pete Weiss wrote:

>The algorithm of which LISTSERV(R) hosts to use for any particular
>distribution CAN be influenced by how many subscribers at a site (or
>possibly domain) are receiving that particular item.

Thanks for all the suggestions. This one seems to be the actual cause
as there are more than one subscibed addresses on my list from my
ISP. After some experimenting and digging around, it turned out that
if just one address was subscribed, the messages came directly, but
if more than one were subscribed and received the same message, the
messages took the extra hop. That would explain why error reports,
command confirms, etc., come directly, and may also relate to this...

On 12 Dec 98, at 12:20, Bill Gruber wrote:

>A NBNET.NB.CA subscriber complained that she was no longer receiving
>LISTSERV list mail but that she was able to receive other mail.

If she is not the only NBNet subscriber on your list that is
receiving posts, I have a feeling the same thing is happening here
and that her (and the other NBNet subscribers) messages are taking
the extra hop through the upstream university. The university
appeared to experience server problems from late Wednesday
night/early Thursday morning until Friday morning, so I suspect this
is the trouble they informed her that had been fixed, and why other
messages got to her fine. I have a feeling that the extra hop isn't
something that can be "fixed" short of only having one NBNet
subscriber receiving posts at a time. So long as the upstream
university doesn't experience problems very often, it shouldn't cause
a problem that often, or so I hope.

Thanks again,

Patrick