Blade Server

[MG2R]

sudo sucd datacd <user name>ln -s /path/to/my/files "Desktop Documents"

And that's all of it. I'm done, I can now access any of my documents from anywhere!

 

Oh, does OwnCloud finally support just dropping in your files like that? That didn't use to the case.

 

It might not be the best of ideas to run OwnCloud as root, with it being a webservice.

[Blade of Grass]

Oh, does OwnCloud finally support just dropping in your files like that? That didn't use to the case.

 

It might not be the best of ideas to run OwnCloud as root, with it being a webservice.

It seems to be working quite well, so I guess they added it in.

OwnCloud isn't being run as root, it's run under the www-data user, and the directory is only accessible by owner and group (chmod 770, I thought I had added myself to the www-data group, but apparently not).

[Dr. Duck]

Wow, your picture taking skills are almost as good as my pap's, Dr. Duck Sr!

[Blade of Grass]

Storage Update Time!

I got another package in the mail today from NCIX containing two new 3TB Seagate NAS drives, this brings the total raw storage of the server up to 13TB!

Some shots of the lovely drives.

Cable management for this was abit of a pain, sadly I don't have a 6 port SATA power cable, so I had to use two 4 port ones (sadly, the 2 port one was too short to reach the power supply).

Cable management for the drives:

All of the drives together:

Now, because of the way that ZFS is setup, you cannot add additional drives to the array. While this may make sense in a business environment where you buy a bunch of drives at a time, I will probably by slowly adding more drives, so that doesn't work for me.

So, today I'll be moving to mdadm and LVM. Let's get this started.

Installing mdadm and LVM

To start, instll mdadm and LVM by running

You will be presented with a screen like so:

As I will not be booting off of my array, I typed in "none".

Next you get this screen on automatially starting the arrays:

I answered "Yes" as I would like it to start the arrays automatically.

Next, you get a screen asking about monthly redundancy checks:

I answered "No" as I'm going to setup a weekly data scrub, so I don't need mdadm's built in one.

Next, you get a screen talking about the mdadm monitoring daemon:

I answered "Yes" and left it as root (I've made it to it automatically forwards my root mail to a separate email account).

Now that's it for the install of mdadm. LVM installs without any questions.

Settings Up the Drives

Now, before I go ahead and erase the partitions on my old drives, I backed up all my data to my extra drive by running rsync.

To start off, I'm going to be preparing the drives using gdisk. The four drives that I will be using are located at sdb, sdd, sde, sdf. I'm going to partition them using the following command:

And now I repeated for each additional drive until they were all formated the same.

Now, I'm going to create the array:

This created the mdadm array. Now it's time to setup LVM on top:

After creating the array and making the physical volume, the HDD idicator light on my server glowed, as it says it will continue for the next 25 hours.

Just to break down what I'm doing:

Now, I wanted to test the drive array performance, just to see how it was doing, so I used hdpram.

I'm not quite sure what's up, but apparently my RAID array has reads of 13GB/s.

Next I'm going to save the array to the config file so that it recognises it on boot. It doesn't seem to work under sudo so I had to change to the root account to do it.

Finally, I am going to create a filesystem on the logical volume. To do this I am going to use mkfs.ext4

And then I'm going to add an entry into fstab

And that's it! It's all setup. I just need to wait for mdadm to resync all the drives, but I can still start using it.

sudo apt-get update && sudo apt-get mdadm lvm2

sudo gdisk /dev/sdbn<enter><enter><enter>fd00 # This is the Linux RAID filesystemwy

sudo mdadm --create --verbose /dev/md0 --chunk=128 --level=6 --raid-devices=4 /dev/sdb /dev/sdd /dev/sde /dev/sdf

sudo pvcreate /dev/md0sudo vgcreate VolGroupArray /dev/md0sudo lvcreate -l +100%FREE VolGroupArray -n Storage

pvcreate <location of the array>vgcreate <name of colume group> <location of PV>lvcreate  -l <percentage of total space> <name of volume group> -n <name that will appear in /dev/mapper/>

sudo sumdadm -Es | grep md >> /etc/mdadm/mdadm.confexit

sudo mkfs.ext4 /dev/VolGroupArray/Storage

sudo vim /etc/fstab# mdadm Storage Array/dev/VolGroupArray/Storage       /storage       ext4    defaults        0       0

[Blade of Grass]

Mini Mini Update

After 21 or so odd hours, the hard drive activity light has finally faded (meaning mdadm has finished it's resync of the array).

Over the 21 ish hours, the server compared and/or wrote 12TB of data, meaning it averaged at 158MB/s.

I also decided to benchmark the array to see how it performs.

Write performance:

sudo dd if=/dev/zero of=/storage/text.out bs1M count=1024010737418240 bytes (11 GB) copied, 58.4555 s, 184 MB/s

So it performs the same as just under 2 Seagate NAS drives.

Read performance:

dd if=/storage/text.out of=/dev/null bs=1M10737418240 bytes (11 GB) copied, 40.3181 s, 266 MB/s

Maybe abit lower than I expected (as it's abit below the read speed of two Seagate NAS drives), but it's definitely more than usable.