di Franco Lanza
Il vero sistemista e’ un po’ come il meccanico di una volta, quello che se gli portavi la macchina per rifare la convergenza e quando arrivavi sentiva che il minimo non andava bene, ti faceva la convergenza, e giustamente la pagavi, ma poi ti sistemava anche il minimo e non ti chiedeva nulla, lo faceva perche’ non sopportava di sentire una macchina che non era a punto come si deve.
Era quello che da ogni minimo e impercettibile rumore indovinava subito qualsiasi problema, anche quello di cui il cliente non si era ancora accorto.
Era quello che dopo cena a casa con la famiglia, tornava in officina, dove potevi vedere le luci accese fino a notte tarda, perche’ stava lavorando al “suo” gioiello, una qualche macchina semi d’epoca recuperata chissa’ dove che con passione piano piano sistemava fino a farla tornare nuova.
Ecco, il sistemista e’ come quel meccanico, e le sue auto sono i server.
In its own words:
Source code differences between two consecutive versions of the Security.framework, a MacOS/iOS component. The seemingly innocuous extra
goto fail; is the cause of a severe security flaw in most Apple products.
Because I have been busy doing this:
- Add SID APT source by creating
/etc/apt/sources.list.d/sid.list(use your nearest mirror):
deb http://ftp.us.debian.org/debian/ sid main deb-src http://ftp.us.debian.org/debian/ sid main
- Configure APT pinning by creating
Package: * Pin: release a=unstable Pin-Priority: 50
- Install the latest version of
libv8-184.108.40.206at the time of writing this:
apt-get install libv8-220.127.116.11
- Download the nodejs package sources, dependencies and build them:
cd apt-get source nodejs apt-get build-dep nodejs cd nodejs-* debuild -nc -uc
- If you encounter
build-dependencyerrors, you should try first to lower the dependency in
debian/control, both in
debuild. If the build fails (e.g. with
undefined reference to `ev_run') the previous version is missing required functions. So, you must install the updated versions of the required dependencies (e.g.
apt-get install name=versione.g.
libev4=1:4.11-1. I suggest this because you’ll have to manually update packages installed from
sid, so the lesser, the best.
- Install the generated package
dpkg -i nodejs_*.deb nodejs-dev*.deb
- Profit :-)
neo-tux by sagarkshetri.com.np
So you have a Linux VM you use for development, because you want to mirror the production environment as closely as possible. You have many applications to deal with, they have to be running at the same time because they are nifty REST JSON web services.
You are very tired to remember which one you put on port
8081, and your configuration files slowly become a real mess. So you set up IP address aliases in for the network interface and decide to assign even host names –
/etc/hosts is just fine – for each app.
Then, in such a setup, why would you still need to run them on ports higher than
1024? Wouldn’t be just great to type the application name in the browser address bar? Indeed it is, but it’s better to not run them as
The solution are Linux capabilities (see also here). The one that interests us is
cap_net_bind_service: it gives a process the right to bind well-known ports (
< 1024). If you use an interpreted language, of course you’ll have to add the capability to the interpreter itself. That’s why there’s development in the title of this article – you should not set this up on a production server, if you don’t know what you are doing.
One final quirk: if you happen to
dlopen() shared objects that dynamically link towards libraries outside the canonical paths, you cannot load them via
LD_LIBRARY_PATH (e.g. the
SYBASE.sh) as it is ignored for
setcap-ped processes. You should better move the library paths into an
Assuming you are the latest and greatest rails developer, you should become root – or use sudo, as you wish – and
# YOU ARE ON YOUR DEVELOPMENT MACHINE setcap cap_net_bind_service+ep `which ruby`
thin start -a yourapp -p 80 >> Using rack adapter >> Thin web server (v1.2.11 codename Bat-Shit Crazy) >> Maximum connections set to 1024 >> Listening on yourapp:80, CTRL+C to stop ...
“If it is good, they stop making it”, the payoff printed on the conference necklaces, distributed to every participant, along with an über-l33t badge customized with our nickname and the key hash.
Being my first experience at an international security conf (I’ve only been to the ccc2k+7 camp), and being a ph outsider ‘cause I never participated to previous editions, the boot keynote held by FX, staffer and frontman, has been enlightening: “you ought to be here!”, he yelled while pointing at the stage, wearing a white shirt with the Phenoelit logo printed on both arms.
“This conference has never started on time”, he continued, “so there was no reason to do that for this last one”. the schedule is straightforward: party, the next days talks from 12.00PM to 7.30PM, then party, and the last days talks from 12.00PM to 5.30PM. definitely a setup well-playing with the available alcohol :-D.
Afterwards, another speaker informed us that the wi-fi access keys we received at the registration allows us to use a 6 APs/3 repeaters beast driven by an OpenBSD box – they want the audience to hack it because, well, “you are the Worst Case Scenario.” :-)
Thanks to @jodosha efforts and praising the former Javaday event, now renamed into codemotion that brought in Rome many Ruby developers from Milan, Padua and other parts of Italy – the first official Ruby Social Club in Rome has been a great success. Of course, officialty is measured only in the amount of twitter spam posted about it! :-): earlier RSCs in Rome go back in time to 2006 organized by current mikamai members and more meetups promoted by @jeko in 2007.
What matters is that there's a community, there's a passion, and there's love to share knowledge - no matter who holds the meetings, the important thing is that they're being held :-).
The event was simple and direct - some beers first, then my keynote on RVM and Ruby interpreters, then Luca's one announcing his minege.ms project and after real social networking :-). I met @gravityblast after much time we didn't meet, knew the PIP group and met @svarione, @punkmanit, @leonardoperna, @riggasconi @ogeidix and other smart people. Moreover, we spent quite some nice time together, making up a really lousy and funny week-end.
Of course, huge kudos to @nhaima's car - that tirelessly carried us around Rome for two days :-)
Now, looking forward to the next meetup, thanks everyone who participed, who offered me beers and, last but not least, thanks to @etapeta for bringing me in time at the meeting - you're the real hero :-).
I spent the last two days trying to set up the Aluminium Mac Mini (rev. 4,1) as a home NAS server with encrypted storage, and I wanted a BSD system on it. There’s already an embedded OpenBSD onto the soekris gateway, and another companion would have been nice. :-)
Guess what, there’s no way out:
- FreeBSD 8.1 doesn’t complete the boot process, due to a bug in the SATA chipset, NV MCP89;
- FreeBSD 8.2-RC1 boots but, due to the same bug, doesn’t recognize any SATA drive nor any USB umass device;
- NetBSD 5.1 boots fine, handles SATA disks via the generic pciide driver (no DMA, thus quite slow) but, unluckily, doesn’t handle the BCM57762 ethernet controller. I tried with quick and dirty patches to bring the bge driver up to date with -current, but still no luck: the MII link detection works, the card transmits but doesn’t receive. The sdmmc controller as well works with -current but not with 5.1-RELEASE. ACPI works correctly;
- OpenBSD 4.8 boots, can access the SATA drives without DMA, and recognizes the bge network card, but exposes the very same behaviour as NetBSD 5.1 with the -current driver fitted in;
- DragonFlyBSD 2.8.2 doesn’t even enter kernel mode, I suspect due to ACPI bugs;
- PureDarwin didn’t inspire me too much, due to the many blocking issues.
All of them support encrypted storage, I built up a NetBSD CGD disk flawlessly onto dk wedges; FreeBSD has got the interesting gbde(8) and geli(8) GEOM-based tools that I wasn’t able to test, while OpenBSD supports crypto via a softraid personality. Unluckily, support for the, nowadays, exotic Apple hardware is a no-brainer.
So, with no other way left open, I decided to go the Linux route, using the excellent sysresccd, that I elect today as the successor of the pld-linux rescuecd, companion of endless system recoveries :-). Anyway, you’ll need the 2.6.36 kernel to make it boot onto the MacMini4,1, due to the aforementioned MCP89 bug. Ethernet card and SD card reader work out-of-the-box.
Now, I’m playing with LUKS and, while I’m not that competent in cryptography, looks like it is more evolved than the *BSD counterparts, and anyway it is more versatile tool than the tools in OpenBSD and NetBSD. On the latter, having to set up GPT and DK Wedges to make the CGD and synch MBR and Disklabel to make the boot loader work (yuck!), everything coupled with rEFIt is quite a mess™. There’s a GPT loader for NetBSD but I hadn’t a chance to try it out.
I hope this information is useful to anyone who tries a similar adventure, comments are appreciated :-).