Category Archives: email
I am likely to be subject to email hiatus in the immediate future.
On seeing a suddenly-filled default inbox folder (customarily where spam lands, as procmail sorts non-spam), I find an address for me has been used as “From” in what is evidently a big spam run. The unexpected messages are mostly out-of-office auto-replies. A handful are from mailinglists that have been spammed but need “me” to subscribe before “I” can post. Happily my own spam filtering has caught most of the other big class: bounce messages from servers so misconfigured as to accept the spam before identifying it as spam and “returning” it to the victim – me.
After a bit of firefighting to reject the autoresponses and moderate the server load, I instead just deleted the address they’re targeting. Since it’s an address that is publicly advertised, I can’t make that a permanent solution. I shall keep an eye on the mail log and re-enable it when the flood abates. Also to relieve the load on the server, I’ve turned off greylisting. It appears to be OK now, but if necessary I may intervene further.
Interestingly the lists spammed include a lot of my current and former hangouts at w3.org and apache.org. Happily the “from” address isn’t one I’ve used to subscribe to any of those lists, so nothing should’ve sneaked through there as “from” me.
 Or maybe I can. But that’ll be as part of a general revamp of my mail addresses, and needs planning.
OK, mailless is an exaggeration. But I’ve lost a bunch of email addresses without notice. And these are the addresses I give to important people like banks, stockbrokers, share registrars and government agencies, as well as companies I do business with online.
This is basically my .co.uk domain. My public email filters spam quite aggressively, and wouldn’t be smart enough to distinguish between my bank or broker and a spammer. So best to have it reject anything purporting to come from a financial institution, and give real banks/etc a separate unfiltered address as soon as they have a legitimate reason to communicate with me.
The system in place is that I created a new address for each sender. That way, if spam starts arriving at one address I can painlessly delete that address without affecting anyone else. Off the top of my head, I’ve had to do that with amazon@ and johnlewis@, as both those retailers started spamming as soon as they had addresses (probably in violation of data protection law, as I certainly didn’t give them permission to spam – I’m meticulous about reading the smallprint and ensuring “no” boxes are ticked while “yes” boxes are unticked, and not confusing the two).
So what’s just happened, and why is it particularly bad?
When I moved house, I contacted my old ISP to tell them I no longer wanted their ADSL service. I asked about retaining email service (the one that operates these addresses) and they told me they offer that at a cost of £21/year. That sounded good to me, so I said yes please. That was May 3rd (when they also charged a termination fee for the ADSL service), and the mail continued to work as expected until this week.
Then on Monday afternoon I got a flurry of messages in my inbox: all my aliases (the actual addresses) and the mailbox had been deleted. The final one told me my “product change” to their mail service was now complete. WTF? I tried to log in, but my username/password no longer worked. Ouch!
That’s a serious problem. With most of my correspondents, there’s no easy way to tell them of a change without going through a confirmation, which involves them emailing me at the old address. And the old addresses are gone, so I can’t do that. Big hoops to jump through, and each of 30+ organisations will doubtless be different. Unless I can restore the addresses – either permanently or just for long enough to change them, but that decision can wait.
Worse, mail to those addresses will get returned instantly as undeliverable, causing mail systems to mark the address as invalid. That’s completely different to a mailserver just being out of service, when the sender’s mail server will queue it (typically for five days, which probably hasn’t been necessary in practice since about 1990) without bothering the sender. So even if I restore the addresses I may have to jump through some painful hoops to restore communication with some folks.
Especially my new bank, with whom I’m in the process of signing up for online service. I’ve been trying to explain a couple of serious faults in their website, and now I’ve surely lost all credibility with them 😦
Given the hassle of hijacking so many addresses, my first reaction was to restore them. So the first thing I did was to ‘phone the ISP and ask them to deal with it. They admitted to having f***ed up, but all they would do was put my problem in a queue of support requests, to be dealt with in 48 hours! I explained this was both mission-critical and qualitatively very different from a mere server outage. But despite admitting fault, the person had no power to prioritise me.
I’m minded to ask for a substantial sum in compensation. And get a lawyer to repeat my request more forcefully if it’s not forthcoming.
A special case is Apache (ASF) mail, which accounts for the vast majority of the volume of messages through some busy mailinglists, as well as mail to my apache.org address. I was able to reroute that at some minor inconvenience, but because this hit me without notice, an hour or two’s mail from Monday afternoon will have been lost.
Some of my mailinglists have been remarkably quiet of late. And some specific mail I’ve been expecting has … also gone quiet.
That’s the kind of thing you furrow your brow at first time, and get increasingly concerned if it persists. Today I reached a critical point: chased down the problem and sorted it.
Background: I run an IMAP server that manages most of my mail. It’s on the server that receives mail for my most widely-used address, and it collects mail from from other servers with a cron job running fetchmail. One of those servers is apache’s minotaur (people.apache.org), which collects all mail addressed to my @apache.org address. And that had stopped working. Not just an occasional failure (that’s usual), but persistently over several days.
I had already logged in to minotaur, and with a local mailer (alpine) verified that mail was arriving for that address. So it’s the fetchmail that was failing. I logged into the imap server, and tried running fetchmail from the commandline. It timed out trying to connect to minotaur.
OK, that’s a POP3 connection tunneled over ssh. What happens if I log in to minotaur and try to connect to POP3 directly from there?
minotaur% telnet localhost pop3 Trying ::1... telnet: connect to address ::1: Operation timed out Trying 127.0.0.1... Connected to localhost.apache.org.
Right. It’s trying an IPv6 connection. But there’s none to be had, and evidently no TCP response. So it’s timing out like a sticky firewall. It stuck on “Trying ::1…” for a very long time – enough for fetchmail to time out.
Solution: change “localhost” to “127.0.0.1” in my .fetchmailrc. Now to wade through all that newly-collected mail 😦
If you emailed me this afternoon, it’s possible I might have lost it. So if I owe you a reply, it’s not necessarily just me being inefficient. Bug me!
I came back to the ‘puter this afternoon to find my inbox clogged with several hundred bounces from a joe-job. So I added a postfix rule to reject the buggers, and set about a mass delete of what was already there. But I’m on the mac laptop, whose mailer expects me to do things individually, and gets a bit sluggish and unresponsive on a mass-delete. It’s possible I may have overlooked genuine mail somewhere in there.
I did spot two non-bounces in there, one of which was a genuine message. And the stuff in the intray excludes my mailinglist mail, which is auto-sorted into per-list folders on arrival.
I’ve just installed postgrey at webthing. So I’m (experimentally) greylisting incoming mail.
Since people say greylisting is so effective, I’m also dropping the more fallible part of my existing spam filtering as part of the experiment. Blacklist access restrictions and sbl-rbl remain in effect, along with basic SMTP well-formedness checking. But all pattern-matching on message headers and bodies is suspended, as part of the experiment.
If greylisting is as effective as many folks claim for it, I’ll make that a permanent change. Time will tell!
The new solaris box came with Thunderbird installed as a default mailer. It works for mail once I’ve disabled crap like pseudo-HTML composition. At least, while online and its IMAP servers are responding. And so long as you post everything through one SMTP host.
OK, yes, that’s pretty limiting. But the real killer is how it falls about in a ghastly heap when trying to access an IMAP server that’s offline or responding slowly.
I’m used to Apple’s mailer – on the mac laptop – which makes a decent job of it. Once it’s synced with the IMAP server on the (linux) desktop, I can access its local cached copies of my mail, no fuss. And it’s pretty good at syncing up whenever a connection is available.
Thunderbird, by contrast, hangs and refuses to open the offline account when the IMAP server is unavailable. Worse, it pops up error messages about it, to interrupt whatever I’m doing. And worst of all, this morning when I first tried to sync it, the IMAPD was responding slowly because the linux box was running updatedb, and has the slowest of cheapo discs. Instead of syncing in background, thunderbird started grabbing all my desktop’s resources, and made X11 more sluggish than ever the linux box’s own desktop gets when running updatedb. It felt like a forkbomb!
So, I want a better mailer. And here’s the rub: I’ve never run a *X box in the same circumstances, with intermittent availability of an IMAP server. I do most of my mail on the Linux box, but obviously its own dovecot instance is always up when i use that. Going back before the days of the mac laptop, I wasn’t running an imapd, so I didn’t have the issue. Back in the days of dialup, I used fetchmail+local folders, as opposed to fetchmail feeding dovecot, and Pine worked just fine.
So, dear lazyweb, what’s a good mailer for *X with intermittent connectivity?