http://pxeboot.com/pxeboot.n12

Booting Linux via PXE from Windows Deployment Services.

There are several guides on the web that claim to explain how it is done. After combining a few of them, it even does. Up to the point where TFTPing a live filesystem (squashfs) fails with an “access violation” from the WDS TFTP server. I think that is because the initramfs, or whatever does the fetching, puts a leading slash into the path. It would make sense for the server to refuse an absolute path on paranoid grounds.

So, let’s try HTTP instead. As luck would have it, there’s an IIS on the WDS already. Configure virtual directory, quickly check permissions, looking good, reboot VM.

404. Huh? The directory returns 200.

IIS log. There’s the 404; win32-status is 50. “The request is not supported.” Double-huh?

Deep down in ancient genetic memory, something stirs. Doesn’t IIS refuse requests for static files with unknown extensions?

Add MIME type mapping. squashfs -> application/octet-stream. Reboot VM.

Works. Nice.

Now. Why does WDS on 2012R2 complain about

wdsutil /add-device ... /bootprogram:foo/bar/pxelinux.com

and tell me that that option should not be used since 2008R2 and “to use a custom PXE prompt policy, […]"? I guess MS decided to abstract from the pxeboot.com/pxeboot.n12 pair into a UI option for F12 or no F12, dropping (un)support for other people’s PXE loaders by the wayside. But at least there is a /force flag to do it anyway. (Or I could change the platform’s default boot program and chain the WDS loader from PXElinux.)

https://www.amazon.de/dp/B01EX1U092

#FakeNews

Schrödingers Paket. Und das geht an eine Packstation, die eher selten “geschlossen” sein dürfte (und es zumindest heute morgen nicht ist).

https://www.youtube.com/attribution_link?a=23MP7ImCK5c&u=%2Fwatch%3Fv%3DzT80zDZgOWU%26feature%3Dshare

Just at the one-hour mark, we see the Keithley 2450 SMU – the world’s second most expensive touchscreen keyboard.

At the top, of course, is the 2460, beating its lesser sibling by another two k€.

https://www.amazon.de/dp/B074VX6TG9/ref=cm_sw_r_fa_asp_rsJhN.57D5MN8

Ich hoffe, die Nähte sind besser als die Übersetzung …

https://www.tagesschau.de/ausland/hurrikan-irma-143.html

»[…] trifft bei den Inseln der Florida Keys auf amerikanisches Gebiet.«

Liebe tagesschau, Ihr habt mal wieder die US Virgin Islands und Puerto Rico vergessen, die Irma schon längst erlitten haben.

»Schnell! Folgen Sie diesem Fisch!«

Eher nicht. Die ZEIT kommt leider ein bißchen spät; die Witzigkeit von »Zukunftsvorhersagen« durch irgendwelches Getier hat sich in den letzten Jahren doch deutlich erschöpft. Oder was kommt als nächstes? »Dieses Photospektrometer mißt fünfmal am Tag die Farbe unseres Wahlbakterienrasens, und wir tun dann so, als täten wir so, als hätte das irgendwelche Relevanz.«

http://blog.zeit.de/teilchen/2017/09/05/npd-politiker-sagt-arabischen-zahlen-den-kampf-an/

Ach, die Rechten … dümmer geht immer.

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222095

#FreeBSD ports maintainers ought to be required to use poudriere to test their updates before they are accepted. There are few things more annoying than building a set of packages only to discover that one has an unsatisfied dependency that the maintainer did not notice.

When building on the “plain” system, any dependencies installed for any given port remain installed. A later port build can then – usually through autoconf – detect and use them.

In poudriere, each build starts with a “clean” environment, and if a port does not declare a dependency, it does not have the package available.

The Porters’ Handbook currently mentions poudriere, but there is no requirement to use it.

Examples:

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=222095

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=220325

MovieOS.