Der Mouse wrote... Good question, and I'm not sure of the answer. I'll find out. I'm certainly not giving up on retries. I just think it's excessive to allow 7 days of tries, especially when most of those are to addresses that don't exist, or don't bother to keep their mailboxes from filling up to their quota. You're welcome to hack the mailman code and give me the patches. I need some time with other things besides this list for a bit, the past two weeks have been a little excessive. Sorry, I've been a bit grouchy lately. Not something I'm particularly fond of admitting, but I have several customers who I maintain their bulk email systems for. On average, each of them process over 2 million outbound emails per week, all with Sendmail clusters. Hey, at least it's subscription based, so I don't feel so slimy. But my point is, I'm quite adept at handling multiple queues. I'm just a little worn out. "figuring out how to do this" is not the issue. Tell you what... once I get the address list cleaned up so that the subscribers are known good email addresses, I'll set up two stage delivery with a different queue runner for "problematic target mailservers". Then I'll set the max time in queue back to a more "reasonable" level. But I'm sure I'll get lots of future questions about email order. Regards, Jay West

When I joined this list, I had to send a personal e-mail to the moderator explaining my interest in classic computers. Apparently this was discontinued fairly quickly because said moderator got a lot of such messages to read. But that was the first day of the list's existance. I guess now the number of new subscribers per day is a lot less. -tony

Sellam wrote.... I'm not positive, but I don't believe mailman will send an attachment to a subscriber. I can include specific text, but the entire faq is a bit much to stuff in the "welcome to the list" email. But, what I CAN do is include a link to the faq in the new subscriber welcome message. Ok, I can do that. But bear in mind that providing for a particular persons mailserver to have an extended outage, means that because of YOUR outage, 14 other people won't get their email for a while. How far do I try to accomodate one person at the expense of others? I'm still considering options.... Jay

On Sat, 5 Mar 2005, Jay West wrote: One old BBS technique is to put a trick answer into the FAQ, marked obviously like "THERE WILL BE A TRICK QUESTION LATER. REMEMBER THIS ANSWER: FOO". Thenk ask it.

On Sat, 5 Mar 2005, Jim Leonard wrote: I concur, this may turn out to be a mistake. There are all sorts of reasonable reasons mailservers can be or appear to be down for hours/days. What precisely is the problem you're addressing? I assume it's queue size and cpu cycles?

I work part-time for a local ISP. One of the things I've been working with them on is their email infrastructure. Interestingly (from the perspective of this conversation), one of the things they have been considering is dropping their secondary MX. They see the maintenance cost as higher than the redundancy benefit, and honestly, I don't think they're wrong. I've never before seen anything but spamware that gives up as soon as the scheme you outlined. Which != more than one mailhost. My own, for example, has two MX records, but they both end up at the same host. (I'll be happy to explain why - ask me off-list....) ...yes? Is...unacceptable? Is...worth bouncing mail addressed to? Feel free to drop me from the list if you really feel that I am sinning that egregiously against the email system by running my own mailserver, and running only the one. (I won't _like_ it, but it's your list.) /~\ The ASCII der Mouse \ / Ribbon Campaign X Against HTML mouse at rodents.montreal.qc.ca / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

...which != "out of their house"; for example, my mailserver is in my condo, but addressed out of a statically assigned /28 - which is SWIPped and has rDNS delegated to me. So far this hasn't proven to be a problem. Averaging over the last 30 days, I have used 3158.7 bytes/sec incoming, 2235.1 bytes/sec outgoing (both figures in error by less than 1 unit in the last place shown, assuming netstat is telling the truth about byte counts). Certainly my DSLAM could in principle be maxed out, but that seems unlikely; more likely would be that my wire would be maxed out, something that doesn't seem to happen in practice - at least not unexpectedly. (I have for example deliberately redlined it when stress-testing it or the networks adjacent to it; I don't count that.) Actually, for me, I believe it is the best way to run a mailserver. That is, of course, taking into account issues such as cost. Yes, if cost were not an issue, I agree there would be ways in which it could be run better. But I daresay the same could be said of almost any mailserver in existence. Perhaps - but my experience over the past few years has been that I have fewer problems inbound to my mailserver than through other mailservers which forward to me. Now, that could be just a question of what I view as "problems", and where I see the problem as lying when I see a problem as present. But since it's my mailserver, it's my metric of goodness too. /~\ The ASCII der Mouse \ / Ribbon Campaign X Against HTML mouse at rodents.montreal.qc.ca / \ Email! 7D C8 61 52 5D E7 2D 39 4E F1 31 3E E8 B3 27 4B

On Sat, 5 Mar 2005, Jay West wrote: Having a data center to play with makes you spoiled :) I've run my own mail server on the same little PentiumPro system for at least 6-7 years now without many problems. In fact, it runs pretty much problem free. No need for redundant mail servers. And I certainly don't want to trust my mail to a third party. I like controlling my stuff...that way I can be sure that if it gets f-d up, it's my fault and not the fault of some moron whom I then have to rely upon to fix it. -- Sellam Ismail Vintage Computer Festival ------------------------------------------------------------------------------ International Man of Intrigue and Danger http://www.vintage.org [ Old computing resources for business || Buy/Sell/Trade Vintage Computers ] [ and academia at www.VintageTech.com || at http://marketplace.vintage.org ]