Un collègue ne locke pas son poste de travail, et vous voulez lui faire une blague ? Un journal linuxfr propose quelques idées. Les plus fines :

• Ctrl+s dans ses terminaux (mais il faut tomber sur quelqu’un qui ne connait pas)
• setxkbmap dvorak (ou setxkbmap fr, selon le collègue !)
• alias ssh=logout

Au passage, j’ai aussi découvert cowsay.

qemu’s documentation isn’t really crystal clear, and I lost a lot of time with this today.

The way you can link your guest system and your host system has changed between qemu 0.7 and 0.8, with the introduction of Virtual LANs. They are mandatory: you always have to use one, even if you only want to use one qemu instance.

When running qemu, you have to specify both endpoints :

• the one to the host system (using -net tap)
• the one to the guest system (using -net nic)

A working command line to start qemu looks like qemu -net nic -net tap -hda ….

Sometimes, you are at an airport, hotel, train station, … Wireless Internet access is available, but very expensive, and you start to feel not so well because you haven’t checked your mail in the last 6 hours.

Often, DNS works : you can resolve everything, which is quite frustrating… But wait. Why can’t we tunnel IP packets into DNS packets ?

There are two existing solutions (to my knowledge) to do that. Both work using the same principle : you delegate a subdomain to a special DNS server (so you need to be able to run a DNS server somewhere, that means having UDP port 53 still available). Then, the client software issues DNS requests for TXT records in this subdomain. Data is encapsulated into those DNS requests.

NSTX (latest version, project page, Debian package) is written in C and tries to build a full VPN between you and the server. While it works, it doesn’t always work, probably because of some caching DNS servers which might not like your crafted packets. Also, it requires setting tun interfaces at both ends. So I gave up with it, and switched to OzymanDNS

OzymanDNS (latest release, no real documentation available) is a simpler alternative, written in Perl. It only allows you to tunnel a TCP connection, which is enough, because if you can establish an SSH connection, you can use SSH tunnelling, Socks proxying, or even build a full VPN with OpenSSH 4.3.

To run OzymanDNS, you need to install the necessary perl modules (Debian packages libnet-dns-perl and libmime-base32-perl, amongst others) on both ends. You also need to delegate a subdomain to your crafted server (we will use tunnel.example.com). Then :

• On the server, start: ./nomde.pl -i 127.0.0.1 tunnel.example.com (the -i 127.0.0.1 is not important, it’s just the IP address the server returns in case it receives an A query).
• On the client, run ssh with ProxyCommand: ssh -o ProxyCommand="./droute.pl sshdns.tunnel.example.com" localhost. The sshdns asks the server for a connection to localhost:22.
• You know have an SSH connection open to the other end.

While it’s not very fast, it’s still quite usable for stuff like reading mails with mutt. Also, I experienced some crashes while trying to transfer a quite large file with scp over socks over ssh over IP over DNS, so there might be some bugs. If somebody wants to hack on this … :-)

Update: Nice presentation about all this stuff here.

Update 2: As mentionned in the comments, there’s also iodine, which is similar to nstx (tunnelling IP over DNS using tun devices on both ends). I also found DNScat. It’s written in Java, and can do tunnelling using ppp.

Je me demandais ce qu’il existait comme bonnes ressources pour apprendre à écrire des scripts bash. Je suis tombé sur linuxcommand.org qui est pas mal du tout, et qui en plus a une bonne page de liens. A garder sous le coude.

dd is a very frustrating application. You often run it through complex pipes (well, I do), and you never know if it’s performing well (or if you forgot to tune the HD perf with hdparm before). The only thing you can is wait.

Well, not quite. If you read the doc, you would have found this :

       Sending  a  USR1  signal  to  a running ‘dd’ process makes it print I/O        statistics to standard error and then resume copying.               $ dd if=/dev/zero of=/dev/null& pid=$!               $ kill -USR1 $pid; sleep 1; kill $pid               18335302+0 records in 18335302+0 records  out  9387674624  bytes               (9.4 GB) copied, 34.6279 seconds, 271 MB/s

Interesting, isn’t it ? But I’ve found something even more interesting : pv (packaged in Debian/Ubuntu).

Description: Shell pipeline element to meter data passing through  pv (Pipe Viewer) can be inserted into any normal pipeline between two processes  to give a visual indication of how quickly data is passing through, how long it  has taken, how near to completion it is, and an estimate of how long it will be  until completion.  .  To use it, insert it in a pipeline between two processes, with the appropriate  options. Its standard input will be passed through to its standard output and  progress will be shown on standard error.

It’s very easy to use :

# dd if=/dev/hda12 | pv |dd of=/dev/hda13 88.2MB 0:00:04 [23.3MB/s] [                                               ] 

And it can even give an ETA if you use it like cat (e.g: pv file | nc -w 1 somewhere.com 3000).