In the end, I resolved this issue using one of several tools found online. Being a QMail novice, there are several things I could have done differently from the beginning that would have saved me several hours.
Due to an unrelated issue (I'm cheap, and don't want to pay for more backup space), I needed to reduce the volume of data on my server. Upon exploration, 1.6G were sitting in "/var/qmail/queue/mess". A quick peek inside indicated that the message sitting in the queue was a failed delivery notification regarding to the postmaster. Unfortunately, postmaster is the failed location. This appeared to be causing a continous loop of notifications where the postmaster kept failing to send emails to itself, and sent a message to itself (with the entire previous emails appended) to notify of the problem. This email had grown to extremely large size.
The first step was to create a valid location for the email to be delivered to (to stop the infinite loop). I created a "Postmaster" account, though it likely would have been better to alias the postmaster to another account.
Instructions on creating a new email account:
http://www.whirlycott.com/phil/pop3.html
Once I did this, and restarted QMail, I was expecting the queue to magically clear itself. This did not happen (maybe I was impatient), so I decided to clear the queue manually. I did this the wrong way. The correct way is to use a removal utility:
Removing Queued mail:
http://www.linuxmagic.com/opensource.../qmail-remove/
I removed the mail by deleting the queue folder. The problem with this is that the created files are named pipes (I think). They have very specific names for a reason. This meant I could not just create a the files/folders again and have them work. Instead I needed to use a repair utility to put all of the items back to the way they were supposed to be.
Repair utility:
http://pyropus.ca/software/queue-repair/
I did find the documentation a little confusing. In particular, the documentation states that things will be auto-detected but as I worked through the program, it kept telling me that I needed to be more specific. In the end, it only worked once I had specified every option:
Code:
./queue_repair -r -c -b -s 32
The options are clearly explained in the documentation, however I did not know if I had the BigToDo, or not (-b -n options). I tried Big, was wrong, used Not, was correct. I do not know of a good way to tell if you have BigToDo, but if you haven't deleted your "queue/todo" folder then you can see if you have folders in there. If you have folders, you have BigToDo.
Hopefully this will save someone else my pain.
Interestingly, the company I lease my server from has been arguing with me about the usage. I have been burning resources on that machine since I first leased it. They told me to fix the problem (as I was consuming more than my allocated resources), but I had reported it as an issue from the initial install (from the first login).
Currently, my CPU usage has dropped from 144% to 0%, and my logs are not growing at insane speed, and I have shed 1.6G from my HD (33.3%). I can't wait to tell them I fixed their mistake.
