Expecting rDNS is pretty common. Expecting PROPER rDNS, on the other hand, is another thing altogether.

If a machine doesn't have rDNS, then it can't send email to anyone at AOL, for instance. It'd be quite disingenuous to say that people who send email through a machine without rDNS would be surprised if they couldn't contact you.

On the other hand, there are too many ISPs who have rDNS, but broken rDNS (doesn't resolve in the forward direction, uses names which don't belong to them, et cetera). I block email from all connecting machines which have rDNS (or HELO/EHLO strings) which say yahoo.com, hotmail.com, gmail.com, or google.com, which cuts down on a LOT of spam. The real services always have blahblah.something.yahoo.com, for instance.

I also block HELO/EHLO names which don't resolve in DNS, and on my backup MX I also block when the HELO/EHLO doesn't resolve back to the connecting IP. This, IMHO, is much more effective than only rDNS checking. People don't always control their own rDNS, but they damned well better control whether their mail server is lying or not.

The bottom line is this: are you expecting email from just anyone? If so, you can't block it but you can increase its spam score. If you generally correspond with the same people and occasionally start corresponding with someone new, you could take the time when someone new has a broken mail server. This is what I've done for years (with HELO/EHLO) and most people thank me once I explain why it's in their best interest to fix it.

How does one get Linux shell access on BSD servers? ;)

I've been doing my own email for 15 years now, and it's really not that hard to maintain. Sure, if your flavor of GNU/Linux changed significantly every time there's a new version, it's a pain to keep up to date, but I've been using similar configuration files, updated a little now and then, with the same software installed across many servers for ages (sendmail, procmail, milted greylist, imap-uw, cyrus-sasl, Squirrelmail for OCCASIONAL webmail only, et cetera).

Some people like to tinker too much to maintain a constantly running server. For them, self hosting is NOT a good idea. Some people like to run GNU/Linux distros which are too difficult to maintain, and again, self hosting isn't an answer. A simple GNU/Linux distro or some flavor of BSD can be much easier to keep up to date and therefore more secure.

There are two primary reasons why I will NEVER move to an outside email provider. The most important one is that in this day and age your email can be subpoenaed without you ever even knowing and employees of any given service can't always be trusted to not do bad things. I want full, 100% control of my email. And in spite of what other people have written in comments about the fact that email isn't secure end-to-end, the archives are always in my possession. But add TLS and at least you've made it MUCH harder for people to see stuff traveling over the Internet.

The second reason is that almost EVERY service is non-deterministic (if I'm wrong, please tell me). I am tired of people wondering where email is only to find out that some cheesy content-based filter silently dropped their email or something else happened and the likelihood that Google or Yahoo will EVER look in their logs to tell you is practically nil. My filtering is based on servers being legitimate, not based on some arbitrarily determined rules. If something is rejected, there's always a known reason and it is ALWAYS logged.

Again, please correct me if I'm wrong, but this has been my experience to date.

I have a Commodore A2232 seven port serial card in my Amiga 4000 in my datacenter which provides serial consoles to a number of other machines. While other multiport serial cards have RISC processors or large buffers, this card is simply a 3.58 MHz 65CE02 which polls each port and puts incoming characters into its 16k of memory, which the Amiga can access directly. It's a beautiful example of simplicity at work.

If you get everything into a standard (free)Unix spool file, it'll be readable a hundred years from now. After all, what other kind of archive file could you have from twenty years ago which you could easily use today?

Since 802.11b can be faster than many Internet connections (at least in the United States), a dedicated network can be used to bridge two or more networks which can use each other in case of an outage. For instance, my work is physically close to my home. Both places are on cable modems, but since throttling happens at the modem, the speed between the two is limited by the uplink rate of each place. By setting up a wireless bridge, I can communicate between the two at about five times the speed (500k/sec as opposed to about 100k/sec) while leaving the Internet feeds usable for other applications.

Also, if the connection goes down on one network, a simple route command on one of the NAT / routing machines makes everything go through the other network's Internet connection.

In the case of high wireless network density (I can see about twenty wireless networks from my work), you can also use 802.11b hardware on channels that aren't commonly used in the US such as 12 and 13 (Europe) or 14 (Japan).

Perhaps it's not ideal, but slow is better than none.

"Google Introduces Command-Line Tool For Linux"

is about as relevant as saying

"Google Introduces Command-Line Tool For Blue Computers" because blue is your favorite color. Sure, it'll run on blue computers, but it wasn't MADE FOR blue computers. Nor were these tools MADE FOR Linux. They'd have to be written as kernel modules to be made for Linux.

Anyhow, Linux isn't even an OS - it's a kernel. Just try to run Linux sometime without GNU and let me know how that works out for you.

Sure, so-called "tech journalists" think that every UNIX thing in the world is really a Linux thing, and sure, no "tech journalist" will ever properly call the OS GNU/Linux, but Slashdot? You people have to be a better example for everyone else.

It was a rhetorical question.

Why not just use Windows?

The reasons some people might give for using GNU/Linux (Linux is just a kernel, after all) are probably similar to the reason that people might give for using MorphOS. Some people like the development environment (especially people who learned on AmigaOS), some people probably enjoy the efficiency of this OS on PowerPC hardware, and some people like to be different and not run the same software as everyone else.

There are no GUIs for GNU/Linux which are as efficient or as intuitive as AmigaOS, and MorphOS tries to continue that philosophy.

If Microsoft OSes are responsible for 99% of all infections, then if they pay for 99% of the tax, that'd work.

Hello,

I have a question about the results of asking a question. I administer a few Ubuntu VMs and I want to simply turn off screen blanking (please note I didn't say anything about running X). How does one simply turn off screen blanking with regards to the standard text login window? Note that setenv and friends aren't the answer because I want screen blanking off always, not just when someone is logged in.

But this isn't about that question specifically. While I still want the question to be answered, what I feel is a more important point is this: the answer is horribly difficult to find. Extensive Googling hasn't provided one. Posting on the Ubuntu forums hasn't provided any help. man pages and digging around in configuration files hasn't demystified anything. This is just an example (but I think it's a good example) where someone decided, "hey, this might be a good idea", but never documented it anywhere nor discussed it publicly.

This, I think, is a growing problem with GNU/Linux distributions. While each attempts to make things easier for the casual Windows convert, the overall cohesiveness of each distro diminishes.

Do you see this as a problem? Do you plan to make changes to the way decisions about Ubuntu are made and, just as importantly, documented? Do you plan to make Ubuntu more cohesive and better organized?

Thanks,
John Klos

This was already posted:

http://mobile.slashdot.org/story/09/10/28/1758226/The-Software-Router-As-MiFi-Killer

Plus, this is obviously a sales ploy to get people interested in the "Connectify" product.

This is probably a plant from someone connected with Connectify.

Just about every machine has some way to do NAT. The first time I ever did NAT was via an Amiga under AmigaDOS 3.1.

Why is this discussion worthy?

NetBSD can be compiled with no FPU calls so that FPU emulation isn't necessary, and therefore defective LC040s can be used. I posted a set a few weeks ago:

http://mail-index.netbsd.org/port-mac68k/2009/10/13/msg000244.html

But to answer your question, yes, I have a full m68040 in my Quadra 605. The 250 gig hard drive is a laptop SATA drive in an Acard SATA-SCSI enclosure (ARS 2000SU). They're not cheap, but I suppose I was a bit sentimental, too. The only limitation that the ROMs and OS have is that the boot partition on a drive must be HFS, not HFS+, but you can run Mac OS 8.1 and use any size SCSI drive up to a full two terabytes. Or, as in my setup, you can use a small boot partition and the rest for NetBSD.

The memory is easy - 128 meg 72 pin SIMMs aren't very expensive. You can find them on eBay if you look for memory for Cobalt Raq2 systems. Since the Cobalt is a 1U machine, the SIMMs won't be very large and will fit just fine.

Here are some of the machines I run:

Cobalt Raq2 with 250 MHz MIPS processor, 256 megs of memory, and 500 gig hard drive - about 30 watts.
Mac mini, 2.26 GHz Core 2 Duo, 4 gigs, 500 gig hard drive - about 20 watts when doing stuff (it maxes out at about 35 watts when the CPUs are pegged).
Quadra 605 (yes, I like classics), 40 MHz m68040, 132 megs of memory, 250 gig hard drive, about 22 watts.

If you want REALLY low power, pick up a Jornada 728 or the likes. It takes THREE watts (the meter shows four watts momentarily now and then), with a 206 MHz StrongARM, 64 megs of memory, and a 16 gig CF card. However, this doesn't help if you want to do lots of file hosting, but I figured it was worth a mention because it takes so little power. I run several Jornadas (including the 690, which uses SH3 processor) as DNS servers running BIND.

The one that might match your requirements best, though, is a Plextor PX-EH25L and the like. You can put in whatever size hard drive you like, and if you want, you can even get the new low power 5900 RPM drives which take half the power of a typical hard drive. It has 64 megs of memory, a 266 MHz SH4 processor, and two USB 2 ports which can be used with a CD or DVD drive, a second ethernet, more storage, or whatever you want to connect. With an inefficient 7200 RPM drive, it takes less than 20 watts (15 to 18), plus it is incredibly small and so far it's been completely stable.

Note that on all of these machines I run NetBSD because I prefer having one consistent OS across all of my architectures, but if you don't mind maintaining a different version of GNU/Linux for whichever you get, you'll be happy with any of the lower power devices.