Send As SMS

Wednesday, August 30, 2006

Looking for a DVD robot

I'm interested in adding a DVD robot to my collection of hardware, not because I want a jukebox but because I'd like to be able to have my automated backup process extend to the actual writing of DVDs. Here's a summary of I've discovered so far.

Firstly, the point of offline (as opposed to offsite and/or merely off-spindle) backups for me is protection from:
  • Bugs in software (OS or application) causing data damage which goes undetected for months.
  • Damage to data by malware.
  • Damage to data by an intruder.
  • To some extent, damage to data through error by an administrator (me!).
The fourth point is ultimately rather difficult to overcome (if I go mad, wipe all of my primary media and make a bonfire of all of my backup media, well, then the data's probably gone), but the proceeding three can all be addressed in a completely reliable fashion through the use of a duplicator robot that cannot reload disks once it has written to and ejected them. Granted, a robot which can retrieve disks once written is an unlikely victim for all but the rather unlikely case of the very determined intruder, but it would be nice to have. Futher, this could readily be combined with a requirement for unattended offsite backups by deploying the robot with a supply of DVDs, perhaps 20. Allowing 2-3 disks per month, such a device could be left alone for months on end.

So, such devices are readily available, the MicroBoards GX-1 for example, but they're a little pricey (£1,637.95 at the time of writing). An interesting additional feature of that sort of device is that, as they're intended for duplication, they often include a printer. This is hardly essential in my application, but if I were to return to the device after leaving it alone for months and it had 10-20 unlabelled disks in its output hopper, then I'd have some manual fiddling (labelling and confirming) to do; a printer is an appealing feature.

A considerably cheaper device (no printer and, by the look of it, an output hopper which the robot would be able to retrieve from) appears to be known variously as:



  • MF Digital's Baxter (£571.58 from cd-writer.com), the response to the question about Linux support was terse: "Raz - Baxter is a windows based device - it doesn't work with Linux."


  • Disc Makers' Pico (US$699.00), I've not asked them about Linux support.


  • Ripfactory's MT1 (emailled response to my enquiry: £449 plus VAT and the shipping to mainland UK is £13 plus VAT), their rather helpful response on Linux support was:

    "The SDK is available for an additional fee of £99 and comes as standard as a windows SDK plus Driver but has Linux drivers available - the SDK is a C and C++ workspace and is simple to integrate.

    In use, the actual drive is read as an external USB device and can be written to as such with the robotics section taking the only work for integration."



I suspect that the robot is being OEM'd, would love to get closer to its source (and corresponding price) and can't help wondering whether the robot will respond to normal SCSI "change media" commands (meaning that the SDK is not required). In any event, it outside my immediate budget, so I'll leave it for the time being.

Tuesday, August 29, 2006

Sneaky DNA analysis to be outlawed

From New Scientist:
Genetic trophy hunters, beware. From Friday next week it will be illegal in the UK to covertly analyse someone's DNA.
...
part of the Human Tissue Act 2004
I can't help thinking that illegality won't be enough to stop this from happening. Clearly it will slow tabloid journalists looking for dirt on celebrities (to publish the results would require publishing an effective confession of having performed, or at least profited by, a criminal act; but who knows, perhaps concerns for "the public interest" may be permitted to override this) and it might deter some misguided employers who might otherwise be tempted, but as this can be done so easily, covertly and fairly inexpensively, I suspect that people who believe that there's some benefit in it are likely to do it anyway.

Cue Gattaca's gene-testing booth which helps young ladies (who have covertly obtained samples of their date's DNA) decide whether the genetic material provides an adequate basis for developing a relationship further.

Friday, August 25, 2006

Sending big files via SMTP, or, is there anything that socat cannot do?

I had a need to copy a large file from one place to another and it happened that an existing SMTP server at the destination was an ideal way to receive it, which left only the question of how to send it.

The file is huge (tens of MB) which more or less precludes using most mail clients to send it and, anyway, I had no desire to have it forwarded by an intermediate MTA which might choke on it, bounce it, or delay it. So, I started poking around at what APT knew about which might be of use.
  • Ssmtp and nullmailer are both candidates, but both require messing about with config files, and installing either of them will conflict-out any existing MTA.
  • Fetchmail appealed, as it uses SMTP as its primary delivery mechanism and allows you to specify the host to talk to but, as far as I can tell, offers no means to process a message file that you already have in your possession; it's raison d'etre is to fetch mail from elsewhere and, consequently, it requires a POP/IMAP/etc. server to provide it with the message.
  • Procmail (+formail) seemed like another approach but, as far as I can tell, it will only send via the /usr/lib/sendmail of the installed MTA, which is exactly what I did not want to do.
  • While peering at the formail documentation, I mused that this would be a cool addon to socat. On a whim, I checked the man page for the string SMTP and, naturally, it was there!
So (assuming Debian):


# apt-get install mpack socat

$ mpack -s "BigFile attached" -o message.txt BigFile
$ socat EXEC:"/usr/share/doc/socat/examples/mail.sh -f sender@example.com recipient@example.com",fdin=3,fdout=4 TCP4:mail.example.com:25,crnl <message.txt

Notes:
  • Mpack takes care of creating a valid RFC2822 message containing the file (split into chunks if required) to deliver via SMTP in the first place.
  • I've modified the manpage example to provide the full path and remove bind= and mss=; they are interesting options, but rarely required.
  • I've also modified the manpage example to add the undocumented '-f sender' option to mail.sh, which matters if the SMTP server is validating sender addresses and $USER@$(hostname) happens not to be a valid email address.

Friday, August 18, 2006

So convenient a thing it is to be a reasonable creature, since it enables one to find or make a reason for everything one has a mind to do.

So wrote Ben Franklin on his decision to eat some fish despite being a vegetarian...

This quote appears in chapter 4 of his autobiography, which I'm reading - and enjoying - at present. His outlook is interesting.

Tuesday, August 15, 2006

The "Managerial Assumption"

I have noticed for a long time, in many contexts, a blind and frequently incorrect faith that better efforts will lead to better outcomes. I refer to this as the "Managerial Assumption". There are of course a great many managerial assumptions in the world, but this is the one that justifies managerial activities to begin with.

That this assumption is not always true is apparent in a number of contexts:
  • Micromanaging people rarely achieves better results, and usually achieves worse results, than a more mature approach or even than leaving said people to their own devices.
  • The now essentially settled argument between datagram routing and circuit switching; the idea that having intelligence throughout the network making each little piece of the communication path "better" would of course lead to better results. Consequently IP was initially ignored, and then later opposed by telcos who were entrenched in their circuit-switching "control [manage] everything" mentality. Needless to say, the recent stampede of telcos towards IP suggests that they've at least noticed a problem with the "better efforts lead to better outcomes" mentality.
  • Continued Australian Government ownership of Telstra in which it is assumed both that more control will lead to better outcomes (than, say, a market-driven approach) and of course the more preposterous idea that government ownership gives "control" of Telstra in any real sense.
  • "Wars" on drugs, terrorism, poverty, etc., in which it is assumed that the problem can be made to go away through better control.
Interestingly, the first time that I understood this idea was not in a managerial context at all, but in Star Wars; Leia (to Governor Tarkin, just before the destruction of Alderaan): "The more you tighten your grip, Tarkin, the more star systems will slip through your fingers." The idea that exerting more control could, in and of itself, cause one to have less control made quite an impression on me at the time.

In a sense, this is a specialisation of the "if a little is good then more is better" fallacy, but it strikes me as worthy of seperate consideration because many extremely weighty decisions turn on this assumption. An example popped up this morning in Fred Wilson's quoting of comments on one of his own postings:
Thinking there is a 'final solution' to complex problems that have been part of humanity since the dawn of time is what creates these catastrophes in the first place.
Well, yes (the rest of the quote is worth reading too).

Tuesday, August 08, 2006

Coping with mail server abuse

I just read an interesting, if somewhat dated, paper by David Mazières on running nym.alias.net. His most interesting contention was that attempts to silence the service by abusing it were actually a comparable threat to attempts to "out" pseudonym users. Many of the problems still exist for mail servers generally, and several have grown larger, but I was surprised to realise/remember how much of the current abuse was already present, in one form or another, a decade ago.

He has also written lots of other fascinating stuff.

Thursday, August 03, 2006

The Great British Beer Festival

Last night I visited the The Great British Beer Festival with Anand.

While I am generally more partial to wine than to beer, the festival was interesting. Amongst the more interesting beers were those with intriguing names like Mighty Oak's Oscar Wilde Mild, those with peculiar ingredients, like Drei Horne's Bananatana, and those with a bit of both like Theakston's Old Peculiar, which is delicious.