Wednesday, February 28, 2007

Spammer buggy software

I just saw these on the subjects page. Seems some spammers script isn't running right - instead of grabbing random words likes it supposed to, its just sending the command it was going to use.

15m10s 21 RANDLINE[./content/words1.txt]
14m3s 19 RANDLINE[./content/words1.txt] RANDLINE[./content/words2.txt]

Hehe. Oh well.. no pump and dump today guys.

Tuesday, February 20, 2007

Received over 5million emails today

It is at 5,178,247 at the moment to be exact. Thats 59 a second. I'm really starting to wonder what the limit is (I'm sure some DOS'r will let me know someday :) ).

The eensy little single Athlon 2100 server is *still* not breaking a sweat.

EDIT: Yeah, that was a short lived record - one day later, we're at 6.1million

Sunday, February 18, 2007

Incoming nasty subject page up & running

As I mentioned before, its at http://www.mailinator.com/subjects.jsp.

I ironed out a few kinks (keeping around spam longer than mailinator wanted to was harder than I thought, the system is pretty draconian about deleting spam).

Anyway - enjoy!

Tuesday, February 13, 2007

What Spam is Mailinator getting - RIGHT Now?

If you read the Mailinator architecture article (post below), you know that Mailinator filters lots of emails in order to handle the millions it gets each day. Now, one way to do that is to filter bulk email. After all, the purpose of Mailinator is to get an email here or there for signing up for stuff or whatever - typically, you don't care about reading the latest in bulk email (as I'm sure you get enough of that in your own email account).

So, if enough emails with the same subject comes in, in a short enough period of time, we stop accepting them. And keep a note that we shouldnt receive these emails anymore. Of course, converting that information from a crufty data structure to a shiny web page is just an hour of hacking away. So, here's the webpage:

Web Page: Banned Subjects








In SystemCountBoxSubject
2h23m26s1019lmyfaqnRE: Marcenaria: Curso Completo com
2h23m23s953oeonlygchlmdyhCialis Shop
2h23m34s924nahvzovamktiViagra Shop
2h23m32s399semsprjskinumsWe are ready to give you a loan
1h2m31s380shadow007Short 30 second form
2h23m26s375barbaraamRefinance approved

This page is in BETA! It will change some. Its also only updated every minute or so.

Note:

Some of these emails get totally deleted (instead of a few initial copies that are entered before we figure out we're about to get about 9 zillion of them). So if you goto a mailbox looking for one of these spams and its empty, its because the email got deleted for other reasons (i.e., its a bounce message, it had a trigger word ("viagra", "schoolgirl", etc) in its title).

Also, the time "In System" indicates when we first saw that message at all. So if it says like 35minutes, then we've been receiving a steady stream of those spams for 35 minutes. Once the steady stream stops (for a few minutes), it will leave the list.

Suggestions welcome!
(I do plan to have a xml feed for this eventually, if you'd make use of that, let me know)

Monday, February 12, 2007

Situation resolved

Not sure if it was exactly a hardware issue. Serverbeach replaced the drive and put the old primary as a secondary. I couldnt mount it until i ran fsck on it. After that, it did have a bunch of lost files but otherwise seems to be working ok.

Anyway, Mailinator is back up and de-spamming!

Sunday, February 11, 2007

WTF - harddrive crash

Harddrive went smush today it seems. Serverbeach is on the job replacing it as we speak (or, I guess, as I type). Should be back up shortly thereafter.

It should be fun to watch the email after its up. Usually, if we go down awhile, a lot of email gets "pent up" and volume is notably higher for a day or two.

Thursday, February 1, 2007

Mailinator's 2006 Stats

I'm definitely a stat junkie and the free Google Analytics definitely feeds my habit. Of course, that's just web stats. For email stats, I've kept my own (far shoddier) statistics. Here's some stats:

Web Stats
The number of hits Mailinator gets is surprisingly consistent. Note the below graph is done with Google Analytics, thus it does not count people on browsers with javascript turned off and it does not count RSS hits (which are several 10's of thousands a day - way more than web hits).



The consistency surprises me because of Mailinator's nature. Personally, I use it "now and then" - like when I need it. Maybe (maybe) once a week when I sign up for something. Its not the type of site that you just hang out and browse on. So interestingly, this "now and then" use spreads across its users in a very uniform way.



What's also interesting is how people find Mailinator. It would seem that people just "know about it". A very large percentage simply come directly to it. A smaller but notable percentage get there through Google. Stumbleupon give a surprising number of referrals (in fact, its 10 times the number of yahoo).

Note that Mailinator's Alexa rating averages around 30,000. I've seen it as good as 19,000 and as bad as 60,000 or so in the last year (seems to bounce a lot).

Email Stats
Sorry this section doesn't have such pretty graphs, like I said, this stuff is all home grown and far less pretty. Also - if you remember (from the architecture entry below) Mailinator's SMTP server is home-grown. That's one reason it can handle this volume but alas it is also a work in progress, thus some stats were lost at times - in other words, the following numbers are estimates although I feel they're still pretty representative.

Number of average emails per day: 1.234 million
Number of total emails for 2006: 450.74 million
Percentage rejected for same subject: 8.7%

Number of total emails for 2005: 280.68 million

Note that "same subject" means mailinator got emails with the exact same subject over and over and over (i.e., bulk email).

Obviously, spam is way way up. And honestly, the surge happened mostly in Q4 of 2006. If 2007 keeps on the track its on now, we'll be headed for (get this) - 1.29 BILLION emails for the year.

How's that for lotsa spam?

Another interesting note is that Mailinator got a total number of web users (not counting those with javascript off and not counting RSS users) of about 1.3million visitors for the year. Say we double that to account for non-javascript and RSS, making it 2.6million for the year.

Assuming that each visitor only cared about 1 email each, that means of the 450 million emails we got, only about .5% were actually looked at. Or 172 or every 173 emails was crap that no one wanted!