Monthly Archives: January 2010

Things can be easy sometimes

remember that I wrote about that KVM wasn’t working a few days ago? Yesterday I finally discovered the problem. I forgot to enable hardware virtualisation in my bios. Sometimes, iam just not so smart :P

mod ruid for apache

Another post about the setup of hostshuis.nl’s servers. This time not about the virtualisation iam using at the server, but about something complete else. A (pretty simple) apache module, but a very nice one (at least, thats my opinion :)). Read more »

Using KVM, its not always working

Yesterda I wrote about using KVM now on 2 servers. Just after I actually wrote that article I discovered KVM wasnt really working on one of these 2 servers (This server was just moved in, and is already a bit older as the other server. Its the old webserver kinda :)).

Read more »

Using KVM for virtualisation, why?

A month ago I wrote already in my previous post that hosthuis.nl is using KVM for there servers. Now that we moved our first server as well to a new datacentre, we are currently installing KVM on that server as well. A lot people think KVM is something complete else as virtualisation, and actually they are both correct as wrong. They are thinking its a KVM Switch, what means you can control serveral server/PC s with 1 keyboard or mouse (Locally, or with a special device via internet). However, this KVM stands for Kernel-based Virtual Machine, something complete else :) Read more »

Greylist stats

Remember that I wrote back in december about greylisting?
Now its nearly a month later, and my main mailserver still uses greylisting. Below here are some stats from this server:

web:~# greylist stats
Statistics since Thu Dec 24 12:59:53 2009 (30 days and 3 hours ago)
——————————————————————-
708 items, matching 12049 requests, are currently whitelisted
0 items, matching 0 requests, are currently blacklisted
332 items, matching 332 requests, are currently greylisted

Of 23386 items that were initially greylisted:
- 708 ( 3.0%) became whitelisted
- 22678 ( 97.0%) expired from the greylist

As you can see the stats are since December 24 last year. Since that moment only 708 pairs of mailserver with emailaddress have been whitelisted (And probarly a lot of that is from a phpBB.com related server :P). Of these whitelisted addresses the server accepted a bit more as 12000 emails.

Nothing special yet. Looking below, since that starting date around 23000 pairs were greylisted. Of these 23000 only those 708 mentioned earlier are whitelisted. That means 97% probarly is spam of what total is received at that server! Maybe there is some real mail in there, but if there is (What I dont think really), that server is probarly not connected properly. I didnt notice iam missing any mail, so I suppose iam not missing anything.

But these numbers are, what I think, pretty good. It means the spamfiltering done at the server level doesnt need to run on that many mails, what means less CPU usage for that. Iam, at this moment, not sure if the RBL checks are done before or after the greylisting, so maybe these numbers can be even higher when iam not using RBLs at the mailserver. And the same applies for some other checks that are done by exim.

Currently, there is one smaller problem with using this method. As backup MX iam currently using the MX from my host. However, when my mailserver now denies the mail for 10 minutes, it goes to the backup MX (Pretty logical ofcourse), however, the backup MX doesnt run a greylisting, so it just accepts the mail, and directly tries to deliver it. However, the greylisting now denies it again, so it needs to wait again 10 minutes. Due to this, the stats are also probarly a bit off.
A solution for this will be to run the backup MX myself, what hopefully will be happening soon, when I get my second box fully installed :P

And now some other real stats. Iam currently receiving at max 10 spam emails a day. Thats something I can live with. But still 10 to much. Luckily enough they are directly filtered and moved to the spam map when they arrive.