When we set up strict overcommit mode , we hadto pick an overcommit ratio, or the alternate way of looking at it, wehad to pick how much total address space commitment we allow. Because wefirst did this for compute servers, we decided to size things so that anactive process would be able to use more or less all of the machine'sphysical memory, plus allow some extra on top to account for systemprocesses that would get pushed into swap.

The logic is relatively straightforward:

• on dedicated

This server recently moved from FedoraCore 2 to FreeBSD, and in the process I discovered that I had quietlybecome addicted to a few vim features, despite past dislikes of overly intelligent things that call themselves vi .

(I am pleased to report that the default FreeBSD version of vim does not behave that way .)

So it turns out to be necessary to set a few vim parameters to getit to behave the way I want. For my future reference, here's what I'vefound I need to

For reasons that do not fit into the margins of this entry, DWiki sometimes needs to get the current Unix load average. I developed DWiki on Linux, and getting the load average on Linux is really easy; you reada line from /proc/loadavg , split it into three floating point numbers,and you're done. So I wrote a get_load function that did that andforgot about the whole thing.

Then this server got changed from Linuxto FreeBSD, and suddenly that code didn't

Since I got asked this in a comment on my entry about clearing metadbreplicas , here is what I remember of how I managed toget a metadb replica so corrupted that it paniced Solaris 10u3 x86.

• I wanted to experiment with metasets on my test machine, so Ineeded a local metadb replica . Because Ididn't know about this I didn't have a spare partition, and becauseI didn't know any better I put the local metadb replica in thattempting slice 8

There is a movement for the default configurations of things like exim to defer sender verification to RCPT TO time;instead of reporting an error or a defer after the MAIL FROM , all MAIL FROM s are accepted and only later does the message start gettingerrors. I have recently come to a realization about why this is wrong,and I even have an example.

The problem is that when you give at least a 4xx error to an RCPT TO ,it makes the sending mailer think that there is a problem with