- Migration Debian from i386 to amd64 without reinstall (Cross Grading)
- N2M: pbuilder for i386 and amd64 on same host
- N2M: parallel build dpkg-buildpackage and pdebuild
- Windows CE: Assertion failed in wincore.cpp line 1034
- CakePHP: Baking a Application in Minutes
- Enabling InnoDB support in debian (MySQL)
The third part stopped with creation of the raid array. As this is a very time consuming process, which takes some resources to complete, this is a good point where to make a break and just let it finish. However, this is not really required, as rebuilding the RAID is also perfectly done by the the OS while you can still use it. But if you stop the procedure – by a reboot, as a example – the kernel will restart the process from the beginning. So take this as a hint, that you maybe want to check the progress before you finally reboot / power down.Just repeating my self, please note, that this VOIDS YOUR WARRANTY. So the usual disclaimer apply: You brick, you repair it. I do not recommend installing linux on the box, if you are not sure you can handle it. Linux requires more hand-on than the GUI -- you won't have a GUI anyway -- but on the other side, it reveals the real power.
Table of Contens
- Part 1: Get Prepared
- Part 2: Installing the Installer
- Part 3: Installing Debian I
- Part 4: Installing Debian II (this article)
- Part 5: Fine tuning after the installation
Back to the text, after your set the RAID up, the partition manager will reload and the next step is to create the „real“ partitions.
After some research, I found out, that my desired filesystem will be xfs or jfs, as these perform very well under system with very limited resources.
However, this was only the plan, as I found out the hard way, that XFS and JFS are currently not working. On the on hand, one can only select „XFS“ in the debian installer, but eventually -- at some later point – the installation will fail:
So I did go for a „traditional“ ext3. Due the troubleshooting, I did not retake all screenshots with that configuration, but the partitioning process is the same except selecting the ext3 as filesystem type. So if you read XFS somewhere, make sure to use the proper file system.
[ WARNING ]DO NOT USE XFS!! THERE ARE REPORTED PROBLEMS WITH XFS ON ARM/ARMEL WHICH CAN CAUSE FILE CORRUPTION! DETAILS: http://lists.debian.org/debian-arm/2008/10/msg00007.html
To create the partitions, you have to select the ones at the top, marked with „RAID1 device#x“. Select them and press enter.
You will be prompted with the following screenshot. Select „Use
As“ to change the filesystem to something more desireable – in
case you do not want a 10GByte swap partition.
After changing the type, you will get some more options, like this (some text might differ). The only important thing is, that you have to set the mount point to „/“, as you want your root filesystem in this partition.
You do that also with the remaining other RAID containers. If you want to encrypt the container which will hold the „NAS“-Data (in my case RAID1 #2) please notice that you cannot do that right out of the installer, but it is not problem at all to set it up later. (Note: Encryption on the Thecus N2100 is quite slow). When you are done, you will have a setup like that:
So, now the time is come to save all the changes to disk and then format the containers to their fs. By the way, the warning is wrong: You have already screwed up your data, and formatting it won't destroy anything as there is currently nothing to be destroyed ;-)
Or short: Say „Yes“.
If you get the following already shown screenshot, I suggest to
read the text again... I repeat it here, as this is where the error
will occur if you try to create the xfs. I did not investigate this
much deeper, as this already took me some hours... This might also be
fixed in some later release of the installer, but as I am not sure if
this is really only a „installer“ issue, I suggest that you
better test a xfs filesystem throughly: I tried after the
installation again to reformat the NAS-Partition to XFS and JFS:
Beside fs corruption and segfaults while repairing. Bottom line it did not work
and I assume there are some problems in the kernel [jx]fs code on the
The popularity contest is some data collection, which helps the debian people a lot. If you are unconfortable with sending some data about your software installation, you should say no here.
After the base system is on the disk, the installer will ask you about what you want to do with your newly installed system. This will help it to select packages to install. Usually you will go for „standard“ or maybe some server selection. „Desktop Environment“ is quite pointless on the Thecus.
But as already said, this is only a coarse selection. I bet you will make your own selection later, using apt-get or aptitude. However, this selection will download and install some packages for you, so again – time for patience.
Eventually this is done, and it is time to flash the „real“ kernel to the non volatile memories. This kernel will then be used to boot debian. The flashing itself is also done by the installer, and this will take several minutes, where a forceful interruption could be fatal and will require the usage of the RedBoot environment. So better wait a little longer before panicking, if nothing seems to happen.
(BTW: Thecus use a different epoch, so the exact date will be bogus. We'll set the clock later when' debian up. )
Installation of the new kernel is almost the last step to be done. Now, you have to select „Continue“ to really finish the installation. By the way, THIS is the point where you might want to check the progress of the RAID rebuilding. Just fire up another ssh session to the installer, and exit to the shell. There you can get the status by „cat /proc/mdstat“ including a estimation when linux think it will be ready.
As you see, the installer disconnected. This is not bad, even if the eye-candy looks so.
All you have to do is to wait some more till the box is up. Then you can ssh in again, using your new root password. At this point, the host identification will – again – have changed. This is normal and ssh's warning can be neglected.
So now it is time to install additional software, update to a recent kernel, etc.
But first, you have to fix the time. Udev and the kernel will create out-of-the box the device „/dev/rtc0“. But as some software will look for „/dev/rtc“, you have to tell udev to create a symlink with these settings:
#add this line to /etc/udev/rules.d/z60_th2100rtc.rules
KERNEL=="rtc0", SYMLINK+="rtc" GROUP="audio"
After that, you might want to set the system clock to the right date. I use the ntpdate package, which I installed with apt-get install ntpdate. In combination with hwclock –systohc the clock can be easily updated. (Note: After the time jumps so badly, the next reboot will cause a full filesystem check. This might take a while, and therefore I suggest to reboot just before going to bed and let it run overnight)
If you want to share your experience with others, feel free to comment here. You can also send me your own stories, and if you want, I can publish them here.
- Installing Debian on the Thecus N2100 -- PART 3 -- Installing Debian
- What NAS?
- La Fonera: A better way to enable RedBoot via Telnet / Ethernet (safely) [HowTo]
- Installing Debian on the Thecus N2100 -- PART 1 -- Preparation for install
- Thecus N2100: Testing a new kernel. Or the Installer. Or Revert. Or Recover.