Bertrand LUPART

Today, i can't start Vagrant, due to a VirtualBox error. It crashes at startup with the following error :

NS_ERROR_FACTORY_NOT_REGISTERED (0x80040154)

Looking on the net, you'll find a load of webpages, mostly related to VirtualBox running on Linux boxen, giving obscure recipes to tweak symlinks and dealing with files permissions.

Those are mostly relevant to Linux systems, and I'd advise not to apply them on your MacOS X computer.

Instead:

• launch Disk Utilities.app
• choose your drive on the left panel
• click on "Repair disk permissions" button

However, impossible to get it working with my two MacOS X 10.5 MacIntel machines, using 802.11n. It just sees the network, tries to connect and complains about some kind of timeout.

Aug 29 09:58:02 welcome System Preferences[581]: Error joining Linkeo: Delai de connexion (-6 timeout connecting)

Hopefully, there was a firmware update for the Linksys WAP4400N to version 1.2.14.

I downloaded a firmware image from Linksys website labelled WAP4400N_Firmware_R1.2.14_01152008.img. It should be corrupted because i never managed to get it recognised by the Linksys device.

Hopefully, there's a WAP4400N_Firmware_R1.2.14.9-ETSI_01202008.img firmware available on linksys-uk website which is usable.

http://www-uk.linksys.com/servlet/Satellite?blobcol

Since the firmware update, my 10.5 802.11n MacIntel work perfectly with the Linksys WAP4400N device.

The interface is web-based, and is known to work only with Internet Explorer under MS-Windows.

Hopefully, if you'd like to configure your linksys switch with MacOS X, http://www.opera.com/ does quite a good job at it.

I just decided it would be time to add that to my 4 interfaces Linux router.

The machine is a Dell PowerEdge 1950 running GNU/Linux Debian etch. I just added a second Intel 1000GT quad port card, so that the machine now sport 10 1Gb/s ethernet ports. The goal was to bond 2 links, using a total of 6 interfaces and keeping 4 of them free.

Adding the first bonding interface was plain easy, as usual. In fact, it was easier than usual because the bonding module were already compiled in, configured and loaded, the bond0 interface already created. In fact, that's what caused me trouble :)

Adding the second one was more problematic: i never used more than one bonding interface to a Linux machine and didn't thought it could be problematic before trying.

While trying to run ifenslave on bond1: i got the following error:

Master 'bond1': Error: handshake with driver failed. Aborting

The error is not quite user-friendly. You really imagine you've now a big big complicated problem. More, when using your friend google, you only find a total of 20 results composed of asian pages you don't understand and… source code :) This really help you to think you're lost.

That's obviously not the case.

Just got to the bonding module configuration again. I've never mentionned how much "max_bonds" i wanted. The default is one, thus, only the bond0 interface was created.

I just added the following at the end of the /etc/modprobe.d/aliases file:

 alias bond0 bonding
 alias bond1 bonding
 alias bond2 bonding
 alias bond3 bonding
 alias bond4 bonding
 options bonding mode=1 miimon=100 downdelay=200 updelay=200 max_bonds=5

Now works as expected.

Don't forget to read Documentation/networking/bonding.txt in the kernel source.

$ pike
Pike v7.6 release 104 running Hilfe v3.5 (Incremental Pike Frontend)
> (string)filter((array)Unicode.normalize("Å i åa ä e ö.", "DK"), `<, 256);
(1) Result: "A i aa a e o."

Original thread on the http://pike.ida.liu.se/forums/:

http://arcknowledge.com/gmane.comp.lang.pike.user/2006-11/msg00016.html

Did you ever noticed some similarities in Coldplay's Talk and Kraftwerk's Computer Love?

http://partyben.com/ did. Check out his http://www.partyben123.com/PartyBen-ComputerTalk.mp3 bootleg, it's really nice.

kernel: martian source 6827efd8 for f738d456, dev ppp0

Some comments found on the net stated these were MAC addresses, while it appears those was too short. More, i couldn't find something that look like that in our servers MAC addresses.

The light came after installing Pike's http://modules.gotpike.org/module_info.html?module_id using http://modules.gotpike.org/docs/UsingMonger.html.

$ pike
Pike v7.6 release 86 running Hilfe v3.5 (Incremental Pike Frontend)
> IP.v4.Address(0x6827efd8);
(1) Result: IP.v4.Address("104.39.239.216")
> IP.v4.Address(0xf738d456);
(2) Result: IP.v4.Address("247.56.212.86")

It's now more clear that my router bothers about one of Google's IP (216.239.39.104) talking to my ppp0 IP (86.212.56.247) on the ppp0 interface.

Now, i can make more tests using ip route get pointed out by redax on #pike:

# ip route get 247.56.212.86 from 216.239.39.104 iif ppp0
RTNETLINK answers: Invalid argument

Fixing this is a different story, but i can now name what's wrong.

• download a Bengli font on http://salrc.uchicago.edu/resources/fonts/available/bengali/ or another location;
• put the font in /Library/Fonts/ if you have administrator rights, ~/Library/Fonts otherwise;
• reload your Bengali page;

http://bn.wikipedia.org/

Just in case you mind, i don't read bengali.

Off course, RAID 1 doesn't help in this case.

The reiserfsck man page says you'd better backup your data before proceeding with --rebuild-tree.

Believe them.

Following this advice, i hopefully made a last backup before proceeding. The reiserfs partition was perfectly readable, except some files.

Crawling the web, i learned on multiple sites that until reiserfsck --rebuild-tree hasn't finished it's job, the partition was unusable at all.

This operation really takes a long time. Don't estimate the time remaining on the number on remaining blocks and the blocks/seconds. Mine was stuck for many hours on pass #2, eating as much as CPU as possible.

Waiting for more than 12 hours for repairing a 200GB partition was fruitless: it ended up saying there were not enough space left on the drive. Not only the drive had a powerfailure while writing data, but some people here were uploading lots of data and there were very little space left (around 5MB). That's something like Murphy's law.

The backup was put in place, i strongly believed the original reiserfs partition were lost forever. I better understand now the amount of blocks reserved for root on ext2/ext3 filesystems. I think i'll have a look if this kind of option is available on reiser, too.

I relaunched the check, mostly for educational purpose. Noone were complaining about lack of data nor data corruption on the backup.

It runned all the night, and on the next morning the drive was totally fixed.

OK, reiserfsck --rebuild-tree did it's job. Thanx to all reiserfs crew. It didn't ate data, but it took almost 24 hours to do that, an amount of time i couldn't make my users wait for.

I discovered i'm now too old to live with the fsck adrenalin :) I hereby promise to make stronger backups now.

Better spend some minutes a day backuping and making sure it's backuped than loose many hours of your life and hours/weeks/month/years of others people work on a power/hard drive failure.

In short, the following command:

sudo fink install pike7.6-gtk

enables GTK, GTK.GlArea and GTK.GladeXML.

More, gtkglarea and libglade (and their dependencies...) are now required for building pike7.6 from source.

features:
dynamic loading..... yes
threads............. yes (posix)
signal handler...... wait thread
cdebug.............. yes
rtldebug............ no
dmalloc............. no
mmx................. no (no mmx.h found)
byte code format.... ppc32
module reloc........ no
use machine code.... yes
int type............ long (4 bytes)
float type.......... double (8 bytes) (ieee big endian)
pointer size........ 4 bytes
svalue size......... 12 bytes (2+2+8)

Bz2................. yes (using libbz2)
DVB................. no (dependencies failed)
FFmpeg.............. yes (using libavcodec)
GLUT................ no (dependencies failed)
GTK................. yes
GTK.GlArea.......... yes
GTK.GladeXML........ yes
Gdbm................ yes
Gmp (bignums)....... yes (version 4.1.3, using libgmp)
Gnome............... no (applet-widget.h not found)
Gz.................. yes (libz)
Image............... yes
Image.FreeType...... yes
Image.GIF........... yes
Image.JPEG.......... yes
Image.SVG........... yes
Image.TIFF.......... yes
Image.TTF........... yes
Image.XFace......... yes
Java................ yes (/System/Library/Frameworks/JavaVM.framework/Versions/1
.5.0/../../JavaVM )
Kerberos............ yes
MIME................ yes
Math................ yes
Math.Transforms.FFT. yes
Mird................ no (libmird/mird.h not found)
Mysql............... yes (libmysqlclient)
Nettle.............. yes
Odbc................ yes
Oracle.............. no (dependencies failed)
PDF.PDFlib.......... yes
PDF.Panda........... no (dependencies failed)
Perl................ yes
Postgres............ yes
Regexp.PCRE......... yes (libpcre)
SDL................. yes
SDL_mixer........... no (dependencies failed)
Ssleay.............. no (use --with-ssleay to compile it)
sybase.............. no (dependencies failed)

As usual, build instructions on http://www.gotpike.org/PikeWiki/index.pike?page=Installing%20Pike