[volunteers] SVLUG Mail Working?

[Rick Moen]

Quoting Mark Weisler (mark at weisler-saratoga-ca.us):

> So, again, thanks for fixing the server and the excellent trouble-shooting 
> clues and tips.

Yr. welcome.  BTW, it seems that the reason we weren't seeing your
pending mail in the Exim4 logs is that it hadn't yet been delivered,
judging by the fact that _both_ your original and your re-post showed
up.  As soon as I cleared the RAM logjam and restarted the affected
daemon proceses, your remote MTA was finally able to deliver to
www.svlug.org the original mails.  That's why your mail then appeared
twice.

> Anyway... so now that mail is moving again, what are the implications of this 
> outage? Will this happen again? 

Hell yeah.

> Soon? 

No, probably not.

> Might mail be lost as, apparently, mine was?

It wasn't, as it turns out.  It was just, so to speak, circling in a
holding pattern, waiting for www.svlug.org's MTA to be able to accept
mail again.

> Do we need more memory for the old server?  [Yeah, yeah, I was a 
> manager...  ;.]

No, not really.

My own server gets by extremely reliably with _half_ as much RAM as the
legacy www.svlug.org server has, and the MTA / antispam setup is
remarkably similar, except that the software in question consists of
more-mature versions, with much, much more emphasis on intercepting spam
in the front-end of Exim4, the C-based, fast, small rulesets, before
handing off mail streams that survive those tests to (slow, bloated)
SpamAssassin (spamd).  _And_ those more-mature software versions aren't
wacked-out 2202-era betaware cobbled together by the Crazy Frenchman.

What he did was really cutting-edge -- in 2002.  As an unfortunate
consequence of how he did it (e.g., non-deb-packaged source tarballs
used to build binaries in /var/local), it's in practice completely
unmaintainable.  

And also, I'm not the least bit surprised to see one runaway spamd
incident over a period of the last few years.  We're lucky it hasn't
happened before.

If the server had had 1 GB instead of 512MB, it just would have taken a
bit longer for spamd to drive the machine into RAM-exhaustion, once the
runaway condition arose.

> Does this encourage us to move to the newer server we have?

Hell yeah.