jomeps weblog

After the installation of Debian, it always annoys me that after I have installed vim (my favourite editor), the system still invokes nano as the default editor. This is the case for example when using “crontab -e”, “svn ci” or “visudo”.

You can see the current editor when calling command
$ update-alternatives --display editor
editor - status is auto.
link currently points to /bin/nano
/bin/ed - priority -100
/bin/nano - priority 40
/usr/bin/vim.tiny - priority 10
/usr/bin/mcedit-debian - priority 25
/usr/bin/vim.basic - priority 30
Current `best' version is /bin/nano.


You can see the available alternatives for editor by calling:
jclbs3:~# update-alternatives --list editor
/bin/ed
/bin/nano
/usr/bin/vim.tiny
/usr/bin/mcedit-debian
/usr/bin/vim.basic


And finally, you can set a new default editor by calling:
jclbs3:~# update-alternatives --config editor

Press enter to keep the default[*], or type selection number: 5
Using '/usr/bin/vim.basic' to provide 'editor'.


I have bought a new home server to store all my personal data on a separate system. I did not spend a lot of money, and I am absolutely surprised how great the server has become.

I have now transfer rates of about 100 MB/s download and 60-70 MB/s upload speed!. The server consumes around 70W and less then 1W in standby mode. The components cost were about 220 EUR only (without harddisks)!!

Here are all the details:
CPU: AMD Sempron 140 (can unlock the second core easily in BIOS)
Board: MSI 770-C45
Memory: Kingston ValueRAM DIMM 1 GB DDR3-1333
Graphic Card: Club 3D CGNX-GS846LCI (cheapest possible, as I use only text mode…)
Chassis: Cooler Master Elite 334
Power: Sharkoon RushPower M 400W
Second Chassis Fan: Sharkoon Silent Eagle 1000

Operating System: Debian GNU/Linux stable

I have three 1TB harddrives (Seagate ST31000520AS) installed as a software RAID (RAID level 5).

For the AMD Sempron 140 processor, it was possible with a simple BIOS setting to unlock a second core. It is then displayed as “AMD Athlon(tm) II X2 4400e Processor” with 2.7 GHz and 5417.69 bogomips per core. This is really amazing, as the CPU costs only 36 EUR – including CPU fan

The board has 6 Serial ATA connectors, from which I am using only three at the moment. As you can see, I did not install a DVD drive. It’s not required, as it was easily possible to install Linux from a USB stick.

When I decided for the parts of the server system, I paid attention that it can include up to 6 SATA hard disks. This is possible now, the chassis, power supply and main board support up to 6 drives.
At the moment, with three drives and RAID5, I have a net capacity of 2 TB (real usable capacity is about 1.7 TB, by the way). So, I still have the possibility to increase the capacity up to 5 TB, without having to buy other hardware components.

Again a post about s3cmd.

Based on s3cmd, I have created a small Linux script called s3infos to calculate the overall size of my S3 data and also the monthly costs for it. Strange, but it seems that such an easy tool does not exist yet. I use it since several months and it works perfectly fine for me.

s3infos will first read the list of buckets in your amazon s3 account and will then loop over each of these buckets to determine its size. At the end, the script shows the total size of your account and the monthly costs for the storage, based on 0,150 USD pro GB.

Example output of s3infos:
Determining buckets in your S3 storage...
6 buckets found.

Total size of all 6 buckets: 43417036586 bytes (41405 MB)
Monthly costs for storage: 6.51 USD


I have scheduled the script in a cronjob, so the system sends me once a week an email with the current size of my s3 buckets and the total costs. See the following cron job entry:

# Send S3 info mail:
0 1 * * 0 /home/jomep/bin/s3infos | mail -s "S3 Infos"

Here the script for download:

• s3infos (size: 1.2 kB, downloaded 349 times)
• Note: you need to rename the file from s3infos.txt to s3infos and set file permissions correctly, for example: mv s3infos.txt s3infos; chmod +x s3infos

You probably know Amazon Simple Storage Service (Amazon S3). If not, it is an online storage web service offered by Amazon, providing unlimited storage. The price is very reasonable (0.15 US-$ per gigabyte and per month). They only disadvantage is that it is very difficult to use and to access – if you do not use any tools for doing so!

Under Linux I found that s3cmd suits best to my needs. It is very simple to use. If you call it the first time, you have to enter your key and secret key, and that’s all.

Now, if you call it again with the parameter “ls”, you can list the contents of your web service:

jomep@jcld1:~$ s3cmd ls
2008-11-09 12:52  s3://jomep-7H3LpA11-backups
2009-07-24 23:57  s3://jomep-7H3LpA11-jcld1-svn-backup
2009-07-17 21:48  s3://jomep-7H3LpA11-photos
2009-07-10 23:07  s3://jomep-7H3LpA11-test

To determine the size of a bucket in bytes simply call:

jomep@jcld1:~$ s3cmd du s3://jomep-7H3LpA11-jcld1-svn-backup

3523030246 s3://jomep-7H3LpA11-jcld1-svn-backup/

Or, to get the size in human readable format:

jomep@jcld1:~$ s3cmd du s3://jomep-7H3LpA11-jcld1-svn-backup -H
3G       s3://jomep-7H3LpA11-jcld1-svn-backup/

The best usage is however the sync parameter. This is what I use, to make daily backups of my data:

s3cmd sync /srv/svn s3://jomep-7H3LpA11-jcld1-svn-backup/

If you use parameter “–delete-removed”, all files that exist on the destination but not (anymore) on the source are deleted.
I use Debian GNU/Linux and installed s3cmd by simply installing the package with the same name:

aptitude install s3cmd

The following commands are listed in the man page:

Commands:
  Make bucket
      s3cmd mb s3://BUCKET
  Remove bucket
      s3cmd rb s3://BUCKET
  List objects or buckets
      s3cmd ls [s3://BUCKET[/PREFIX]]
  List all object in all buckets
      s3cmd la
  Put file into bucket
      s3cmd put FILE [FILE...] s3://BUCKET[/PREFIX]
  Get file from bucket
      s3cmd get s3://BUCKET/OBJECT LOCAL_FILE
  Delete file from bucket
      s3cmd del s3://BUCKET/OBJECT
  Synchronize a directory tree to S3
      s3cmd sync LOCAL_DIR s3://BUCKET[/PREFIX] or s3://BUCKET[/PREFIX] LOCAL_DIR
  Disk usage by buckets
      s3cmd du [s3://BUCKET[/PREFIX]]
  Get various information about Buckets or Objects
      s3cmd info s3://BUCKET[/OBJECT]

For further information have a look on the amazon website at http://aws.amazon.com/s3/ or at the wikipedia article on http://en.wikipedia.org/wiki/Amazon_S3.

I found a good article about s3cmd here: http://s3tools.org/s3cmd.