DWiki has a pretty simple way of displaying comments; all comments ona single page, which also has the main entry. There's a lot of moresophisticated comment handling systems, like LiveJournal and Slashdot,and I won't deny that part of why DWiki is simple is that simple is less code . But part of it is deliberate,because it's all about the context.

Specifically, I want people who read comments to have as much contextas possible, which means that I want them to

Both SANs and cluster filesystems have multiple machines talking tomultiple shared disks that all of them can see. The difference is thatSANs are designed for a single machine to talk to a given disk at atime, while cluster filesystems allow multiple machines to talk to thedisk at once.

Which raises the big question: why do you need cluster filesystemsat all? Why can't multiple systems share a single disk without doinganything special?

There's two problems with shared disks: caches and coordinatingupdates. These

Since I just went digging for this the other day , here's what I know about what controlswhat Ethernet devices get named on Red Hat Enteprise (and probablyalso on Fedora, but I haven't looked at my Fedora systems in thislevel of detail).

• if kudzu is enabled, it uses /etc/sysconfig/hwconf to nameeverything. If there is no such file or the data in it doesn'tmatch current reality, various bad things happen.

  (You can probably hand-edit

Here's my problem: I have a bunch of identical 1U servers (SunFireX2100 M2s) with four onboard Ethernet ports, driven by two differentchipsets (two nVidia ones, two Broadcom ones). I want to configure ourRHEL installs so that no matter which physical unit I stuff the systemdisks into, the Ethernet ports come up with consistent names that matchthe ports on the back of the server; eth0 should always be the portlabeled 'port 0' and

Suppose that you have a software RAID array. Suppose further that youhave a disk or two fail spectacularly; they don't just have errors, theygo offline completely.

Naturally, software RAID fails the disks out; you wind up with somethingin /proc/mdstat that looks like this:

md10 : active raid6 sdbd1[12] sdbc1[11] sdbb1[10] sdba1[9]sdaz1[13](F) sday1[

Here is a thought that just crystallized for me:

In operations, crud rains down out of the sky and the sysadmins haveto make it go and keep it going.

In system administration, at least you get to design and build thecrud yourself.

(Corollaries about things you inherit from a previous sysadmin whenyou move into an existing environment are left as an exercise for thereader.)

Making Exim discard things in general is simple: you create an Eximrouter using the redirect driver that blackholes messages, and guardit with a suitable set of conditions. I decided to be safe, so we onlydiscard locally generated bounces of spam that are going to an outsidedomain.

(You can detect locally generated bounces with the condition ' and{{!eq{$interface_port}{25}} {!def:return_path}} '. The easy way todiscard messages is

We have a spam forwarding problem. Specifically, we forward spam,although not through choice; it's a political mandate not to imposefiltering on people, and some of the people who don't filterforward their email elsewhere, and predictable things ensue .

We're dealing with this in three ways:

• we use a different source IP address when forwarding spam-tagged email , splitting good emailtraffic away from the spam so we avoid contaminating the formerwith the latter.

  This is especially important for places where

We got contacted by a user reporting that he couldn't get mail withan address in kuet.ac.bd; our inbound mail gateway was consistentlyrejecting the address as temporarily unresolvable. When I startedlooking into the DNS situation, all sorts of peculiar things startedcrawling out of the woodwork.

• in the root zone, the bd country domain has two servers, dns.bdand slave.bttb.net.
• if you query either of them you get a third server as well,dns.bttb.net