<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<meta name="generator" content="Docutils 0.5: http://docutils.sourceforge.net/" />
<title>Drobo-Utils</title>
<style type="text/css">
/*
This style sheet is a slight modification of voidspace.css.
mods by Peter Silva...
Voidspace.css authors:
:Authors: Ian Bicking, Michael Foord
:Contact: fuzzyman@voidspace.org.uk
:Date: 2005/08/26
:Version: 0.1.0
:Copyright: This stylesheet has been placed in the public domain.
Stylesheet for Docutils.
Based on ``blue_box.css`` by Ian Bicking
and ``html4css1.css`` revision 1.46.
*/
@import url(html4css1.css);
body {
font-family: "Lucida Grande", "Lucida Sans", Verdana, sans-serif;
/* font-family: Arial, sans-serif; */
background-color: #333333;
color: #cccccc;
}
em, i {
/* Typically serif fonts have much nicer italics */
font-family: Times New Roman, Times, serif;
}
a:link {
color: white;
}
a:visited {
color: white;
}
a.target {
color: white;
}
a.toc-backref {
text-decoration: none;
color: white;
}
a.toc-backref:hover {
background-color: inherit;
}
a:hover {
background-color: #cccccc;
}
div.attention, div.caution, div.danger, div.error, div.hint,
div.important, div.note, div.tip, div.warning {
background-color: #cccccc;
padding: 3px;
width: 80%;
}
div.admonition p.admonition-title, div.hint p.admonition-title,
div.important p.admonition-title, div.note p.admonition-title,
div.tip p.admonition-title {
text-align: center;
background-color: #999999;
display: block;
margin: 0;
}
div.attention p.admonition-title, div.caution p.admonition-title,
div.danger p.admonition-title, div.error p.admonition-title,
div.warning p.admonition-title {
color: #cc0000;
font-family: sans-serif;
text-align: center;
background-color: #999999;
display: block;
margin: 0;
}
h1, h2, h3, h4, h5, h6 {
font-family: Helvetica, Arial, sans-serif;
border: thin solid black;
/* This makes the borders rounded on Mozilla, which pleases me */
-moz-border-radius: 8px;
-webkit-border-radius: 8px;
padding: 4px;
background-color: black;
color: white;
width: 70%;
text-align: center;
}
h1 {
border: medium solid black;
color: white;
}
h1 a.toc-backref, h2 a.toc-backref {
color: white;
}
h2 {
border: medium solid black;
color: white;
}
h3, h4, h5, h6 {
color: white;
}
h3 a.toc-backref, h4 a.toc-backref, h5 a.toc-backref,
h6 a.toc-backref {
color: white;
}
h1.title {
text-align: center;
/* background-color: #444499; */
color: #eeeeee;
border: thick solid black;
-moz-border-radius: 20px;
-webkit-border-radius: 20px;
}
table.footnote {
padding-left: 0.5ex;
}
table.citation {
padding-left: 0.5ex
}
pre.literal-block, pre.doctest-block {
background: #cccccc;
border: thin black solid;
padding: 5px;
color: black;
}
.image img { border-style : solid;
border-width : 2px;
}
h1 tt, h2 tt, h3 tt, h4 tt, h5 tt, h6 tt {
font-size: 100%;
}
code, tt {
color: #000066;
}
</style>
</head>
<body>
<div class="document" id="drobo-utils">
<h1 class="title">Drobo-Utils</h1>
<p>Drobo-utils is a set of linux tools to query and manage <a class="reference external" href="http://www.drobo.com">Data Robotics</a>
Drobo storage systems. Fire up drobom view, and it should look
familiar to those who have seen the dashboard on other operating systems.
Drobom view is built on a little python programmer interface which can be installed
on the system and used by other applications as well. For experienced Linux hands,
the rest of the command line interface is provided by other sub-commands of
drobom, and offer the same functionality as the view graphical interface.</p>
<div class="section" id="compatibility-matrix">
<h1>Compatibility Matrix</h1>
<p>If you can get your drobo device to be associated with a /dev/sdX <a class="footnote-reference" href="#genscsi" id="id1">[1]</a>
file, then you will be able to partition, build file systems and
read & write data to it using ordinary system tools (see <a class="reference internal" href="#setup">Setup</a>.)</p>
<p>Drobo-utils accesses drobos via the same files. The software
scans those files, and asking each device if it is a Drobo.
Unfortunately, the way Drobos respond varies, so not all of them respond
in a way that the software understands. Even for the same
device, different physical interconnects may work with different functionality.
There are two levels of access to Drobos: data (being able to read and
write data, and full meaning that the drobo responds to the full
command and control language.</p>
<p>With that in mind, the compatibility matrix of each device vs. the
physical channel is below:</p>
<table border="1" class="docutils">
<colgroup>
<col width="14%" />
<col width="11%" />
<col width="11%" />
<col width="16%" />
<col width="20%" />
<col width="13%" />
<col width="16%" />
</colgroup>
<tbody valign="top">
<tr><td>Model</td>
<td colspan="5">Interface
( Performance MB/s )</td>
<td>Maximum
LUN Size</td>
</tr>
<tr><td>interf</td>
<td>USB</td>
<td>FW</td>
<td>TCP/IP</td>
<td>iSCSI</td>
<td>eSATA</td>
<td>ext3</td>
</tr>
<tr><td>Drobo
(Gen1)</td>
<td>full
(15)</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>2</td>
</tr>
<tr><td>Drobo
(Gen2)</td>
<td>full
(15)</td>
<td>full?
(?)</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>2</td>
</tr>
<tr><td>Drobo
Share</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>data*1
(15)</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>2</td>
</tr>
<tr><td>Drobo
Pro</td>
<td>full
(15)</td>
<td>?
(?)</td>
<td>n/a
(0)</td>
<td>full*2
(80)</td>
<td>n/a
(0)</td>
<td>8-?</td>
</tr>
<tr><td>Drobo
Elite</td>
<td>?
(?)</td>
<td>n/a
(0)</td>
<td>n/a
(0)</td>
<td>?
(?)</td>
<td>n/a
(0)</td>
<td>8?</td>
</tr>
<tr><td><dl class="first last docutils">
<dt>Drobo</dt>
<dd>S</dd>
</dl>
</td>
<td>full
(15)</td>
<td>?
(?)</td>
<td>n/a
(0)</td>
<td>n/a
(?)</td>
<td>data
(0)</td>
<td>8</td>
</tr>
</tbody>
</table>
<pre class="literal-block">
full - Find/Full: drobo-utils will find the device on it's own (auto-discover)
data - works: device functions for data i/o, but Drobo-utils cannot access it for configuration.
n/a - not applicable (device does not have this interface.)
*1 - Droboshare is not usable with drobo-utils on a linux server. If you
install Droboshare Augmented Root File System (<a href="#id7"><span class="problematic" id="id8">DARFS_</span></a>) then one can
run a drobo-utils in line mode on the droboshare itself.
*2 - Will not detect, out of the box, an iSCSI drobo. One must configure
the iSCSI subsystem to obtain a /dev/sdX file. see <a class="reference internal" href="#iscsi-setup">iSCSI Setup</a> for
details on that initial configuration. After that point, Drobo-utils
will function properly over iSCSI.
</pre>
<table class="docutils footnote" frame="void" id="genscsi" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label"><a class="fn-backref" href="#id1">[1]</a></td><td>Linux drivers make disk, cdrom, and other peripherals, look
like SCSI peripherals to applications. Regardless of the physical connection,
it is a normal part of the linux kernel to make the device appear as a
so-called "generic SCSI" one.</td></tr>
</tbody>
</table>
</div>
<div class="section" id="coffee-is-hot">
<h1>COFFEE IS HOT</h1>
<p>People sued a national fast food chain because their <a class="reference external" href="http://en.wikipedia.org/wiki/Liebeck_v._McDonald's_Restaurants">coffee was hot</a>, but did not
have a warning on the cup stating that. For most people, the risk of scalding
should be fairly obvious.</p>
<p>Drobo has made it so much easier to obtain much more secure storage, that some with
little or no professional storage management experience are getting Drobos.
Some have the expectation that Drobo´s, because they allow for disk failures,
replace the need for backups. Sad stories have been told about people putting
data only on the drobo, and no-where else, and then something happens and they
lose the data.</p>
<!-- PLEASE, PLEASE, PLEASE:: Do not store all your data on a Drobo (or any
other single device, from any vendor) with no backups or alternate copies.
Eventually, Something very bad will happen. -->
<p>You need to look at your data and determine the backup/data security strategy.
If you have never done this, or do not know what it means, please consult the
<a class="reference internal" href="#deployment">Deployment</a> section in this page for examples.</p>
</div>
<div class="section" id="installation-easiest">
<h1>INSTALLATION: Easiest</h1>
<p>On Ubuntu 9.10 <a class="footnote-reference" href="#distro" id="id2">[2]</a> or later (or Debian unstable or other debian derived
distributions), drobo-utils is included in the repositories, and installation
from a shell prompt is simply:</p>
<pre class="literal-block">
% sudo apt-get install drobo-utils
</pre>
<p>to run at least the command line utility. Users on servers often want only
command line functionality. On the other hand, to enable the graphical user
interface, one more package must be installed:</p>
<pre class="literal-block">
% sudo apt-get install python-qt4
</pre>
<p>That is the easiest installation method, this method ensures that any packages
required are automatically installed on the system as part of the above
installation. On other distributions, or if the version in the repositories is
too old, more complicated methods might be needed. For all other installation
methods, one must ensure the packages that drobo-utils requires are installed.
These packages are called Dependencies.</p>
<table class="docutils footnote" frame="void" id="distro" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label"><a class="fn-backref" href="#id2">[2]</a></td><td>Drobo-utils is developed for release on the stable version of
Kubuntu at the time it is released. Development started on kubuntu 7.10
and continued to 9.10 at the end of 2009. Any similarly recent distribution
ought to do. The package is accepted into Debian unstable, so all debian
derived distributions (debian, *ubuntu, MEPIS, PCLinux-OS, etc...) should
inherit the package in due course.</td></tr>
</tbody>
</table>
<div class="section" id="dependencies">
<h2>Dependencies</h2>
<p>Before one can install drobo-utils itself, the other packages needed are something
like those below (these examples are ubuntu packages, names may vary on other
distributions):</p>
<pre class="literal-block">
python -- interpreter for python language
parted -- partitioner, usually included with the distro.
</pre>
<p>If using Redhat Enterprise Linux (RHEL, aka. CentOS, Scientific Linux etc...),
which have python 2.4 <a class="footnote-reference" href="#python" id="id3">[3]</a>, then the following are necessary:</p>
<pre class="literal-block">
python-ctypes -- module for C-interface
</pre>
<table class="docutils footnote" frame="void" id="python" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label"><a class="fn-backref" href="#id3">[3]</a></td><td>I'm not sure that python-2.4 will work, for other reasons. the utility is built on
python-2.5 and python-2.6 and it should work on them. python-2.4 is not tested. python-3 will
definitely not work.</td></tr>
</tbody>
</table>
<p>On RPM-based distros (such as Redhat & SuSe), it would more likely be 'yum' instead of
'aptitude' and some of the package names will change. A typical difference is that
packages for developers have the -devel suffix on Redhat derived distributions, instead
of the -dev favoured by debian derived ones.</p>
<p>Here is an example from fedora 7 (courtesy of help4death on the google group):</p>
<pre class="literal-block">
% yum install python
% yum install PyQt4
% yum install python-devel
</pre>
<p>NOTE: if X or QT is missing, it will only disable the GUI. Line mode will
work without issues. The package should work fine on headless servers using
only the command line.</p>
</div>
<div class="section" id="install-from-package">
<h2>Install From Package</h2>
<p>Once dependencies are satisfied, one can install the latest stable package manually.</p>
<p>Point a browser at: <a class="reference external" href="http://sourceforge.net/project/showfiles.php?group_id=222830">http://sourceforge.net/project/showfiles.php?group_id=222830</a></p>
<p>where the most current packages are available. after downloading a .deb, it is simply a matter of:</p>
<pre class="literal-block">
# dpkg -i drobo-utils-<version>.deb
</pre>
<p>done!</p>
<p>Redhat/Fedora users. alien may be used to convert the package. I don't know if it works. Someone to take on RPM packaging would be very welcome! This is a pure python package, so the chances are good that it does work without issue.</p>
<p>There is a <a class="reference external" href="http://olea.org/paquetes-rpm/repoview/drobo-utils.html">Fedora Package</a></p>
</div>
<div class="section" id="install-from-tar-file">
<h2>Install From TAR file</h2>
<p>Assuming the dependencies are installed/satisfied, the package will actually run fine
without being installed in any systemish places. Source code can be directly downloaded
run it explicitly from the directory.</p>
<p>Point a browser at: <a class="reference external" href="http://sourceforge.net/project/showfiles.php?group_id=222830">http://sourceforge.net/project/showfiles.php?group_id=222830</a></p>
<p>download the .tgz preferred, then unpack it:</p>
<pre class="literal-block">
# tar -xzvf drobo-utils-<version>.tgz
# cd drobo-utils-<version>
# ./drobom status
</pre>
<p>for all of the examples in the manual one just needs to prepend './' before drobom.</p>
</div>
<div class="section" id="install-from-git">
<h2>Install from GIT</h2>
<p>When a new model comes out, or the stable version is missing a feature, one may elect to
follow along with the latest development version. installation of <a class="reference external" href="http://www.git-scm.com">git</a>, is necessary,
then use it can be used to get a copy of the source tree:</p>
<pre class="literal-block">
# apt-get install git
# git clone git://drobo-utils.git.sourceforge.net/gitroot/drobo-utils/drobo-utils
# cd droo-utils
# ./drobom status
# git pull
</pre>
<p>This gives a read-only copy of the source code that can be updated with the latest
changes with 'git pull'. One can also select any stable version of drobo-utils by use of
'git branch -r', and 'git checkout'. For details, consult git documentation.</p>
<p>So, one way or another, drobo-utils is installed. The next step is to try it out.</p>
</div>
<div class="section" id="try-out-the-cli">
<h2>Try Out the CLI</h2>
<p>The first item to verify after installation is to invoke the
command line interface (CLI.) and see if something sensible happens...
on my system with a drobo[#sdz] the following happens:</p>
<pre class="literal-block">
% sudo drobom status
/dev/sdz /drobo01 100% full ( ['Red alert', 'Bad disk', 'No redundancy'], 0 )
%
</pre>
<table class="docutils footnote" frame="void" id="sdz" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label">[4]</td><td>in examples, drive always changed to sdz to avoid copy/paste errors.</td></tr>
</tbody>
</table>
<p>very scary, but my drobo is in bad shape right now... normal result is: []
as a status, which means there is nothing wrong. To get all kinds of
information about the drobo, try 'drobom info.' Invocation
without arguments at all which will cause it to print out a list of the
commands available through the command line interface.</p>
</div>
<div class="section" id="try-out-the-gui">
<h2>Try Out the GUI</h2>
<p>Once the command line functionality is verified, and assuming
python-qt4 is installed:</p>
<pre class="literal-block">
# drobom view
</pre>
<p>as root starts a GUI for each drobo attached to a computer.
There are various tabs to allow one to obtain information from the Drobo,
and also change its configuration. For example, one can use the Format
tab of the GUI to partition the device and create a single file system
for a given LUN.</p>
<p>There are two choices to make when setting up a Drobo: file system type, and LUN size.
For a discussion of these choices, please consult: <a class="reference internal" href="#lunsize-fits-all">LUNSIZE fits all</a>
and <a class="reference internal" href="#filesystem-choice">Filesystem Choice</a>.</p>
</div>
</div>
<div class="section" id="manual-setup-drobo-with-linux">
<span id="setup"></span><h1>Manual Setup Drobo with Linux</h1>
<p>This section provides an illustrative example of the most common configuration.
An ext3 file system is built on a Drobo with whatever LUNSIZE is already in place.
The GUI and line modes produce exactly the same result, and simply
execute standard linux partitioning using parted, and the appropriate
file system builder for the type in question. Sample CLI run:</p>
<pre class="literal-block">
PleaseEraseMyData# drobom -d /dev/sdz format ext3
peter@pepino:~/drobo/drobo-utils$ sudo ./drobom format ext3
/dev/sdz - Drobo disk pack 00% full - ([], 0)
preparing a format script for a ext3 file system as you requested
OK, I built the script but nothing is erased yet...
You can have a look at it with: cat /tmp/fmtscript
If you are really sure, go ahead and do: sh /tmp/fmtscript
WARNING: Ready to destroy all your data. Continue? (y/n) n
Phew... You stopped just in time!
peter@pepino:~/drobo/drobo-utils$
# cat /tmp/fmtscript
#!/bin/sh
parted /dev/sdz mklabel gpt
parted /dev/sdz mkpart pri ext3 0 100%
parted /dev/sdz print; sleep 5
mke2fs -j -i 262144 -L Drobo01 -m 0 -O sparse_super,^resize_inode /dev/sdz1
</pre>
<p>The above sets up the drobo as one big partition, with a label that says
it ought to contain an ext3[#mke3fs] file system. For an NTFS file system,
write ´ntfs´ in place of ext3. The next step is to add the file
system into the partition. while parted's are instantaneous, the mke2fs
takes a while, just have a little patience, it´ll be fine.
The ´coffee is hot´ disclaimer is necessary at this point:</p>
<pre class="literal-block">
WARNING: THE FOLLOWING LINES WILL ERASE ALL DATA ON YOUR DROBO!
WARNING: NO, IT WILL NOT ASK ANY QUESTIONS!
WARNING: ASK YOURSELF, before you start: ARE YOU SURE?
WARNING: AFTER THE SECOND LINE, YOU ARE TOAST.
WARNING: BEST TO BACKUP YOUR DATA BEFOREHAND...
sh -x /tmp/fmtscript
</pre>
<p>(For an ntfs file system, use mkntfs -f -L Drobo01 /dev/sdz1
... For ext3, be prepared to wait[#mkext3time]_ )</p>
<table class="docutils footnote" frame="void" id="mke3fs" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label">[5]</td><td>The proper command to build an ext3 file system is mke2fs -j. This
confuses people who wonder why one doesn't use some sort of ext3 mkfs. There isn't one,
an ext3 is an ext2 with a journal.</td></tr>
</tbody>
</table>
<table class="docutils footnote" frame="void" id="mkext3-time" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label">[6]</td><td>mke2fs takes a very long time to run, on the order of ten minutes
per Terabyte. the display format button just turns red while the format
is in progress. Have not determined a method to monitor progress yet from
the GUI yet. other file systems are much more quickly created, so less of
an issue.</td></tr>
</tbody>
</table>
<p>Sample run:</p>
<pre class="literal-block">
root@alu:~# parted -i /dev/sdz
GNU Parted 1.7.1
Using /dev/sdz
Welcome to GNU Parted! Type 'help' to view a list of commands.
(parted) mklabel gpt
(parted) mkpart pri ext2 0 100%
(parted) quit
root@alu:~# fdisk /dev/sdz
GNU Fdisk 1.0
Copyright (C) 1998 - 2006 Free Software Foundation, Inc.
This program is free software, covered by the GNU General Public License.
This program is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A
PARTICULAR PURPOSE. See the GNU General Public License for more details.
Using /dev/sdz
Command (m for help): p
Disk /dev/sdz: 2199 GB, 2199020382720 bytes
255 heads, 63 sectors/track, 267349 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System
/dev/sdz1 1 267350 2147488843 83 Linux
Command (m for help): q
root@alu:~# mke2fs -j -i 262144 -L Drobo01 -m 0 -O sparse_super,^resize_inode /dev/sdz1
mke2fs 1.40.8 (13-Mar-2008)
Filesystem label=Drobo01
OS type: Linux
Block size=4096 (log=2)
Fragment size=4096 (log=2)
8388608 inodes, 536870886 blocks
0 blocks (0.00%) reserved for the super user
First data block=0
16384 block groups
32768 blocks per group, 32768 fragments per group
512 inodes per group
Superblock backups stored on blocks:
32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,
4096000, 7962624, 11239424, 20480000, 23887872, 71663616, 78675968,
102400000, 214990848, 512000000
Writing inode tables: done
Creating journal (32768 blocks): done
Writing superblocks and filesystem accounting information: done
This filesystem will be automatically checked every 26 mounts or
180 days, whichever comes first. Use tune2fs -c or -i to override.
root@alu:~#
root@alu:~# mount /dev/sdz1 /mnt
</pre>
<div class="section" id="id4">
<span id="lunsize-fits-all"></span><h2>LUNSIZE Fits All?</h2>
<p>By default, Drobo creates a separate 'disk' visible to the computer for every 2
Terabytes (TiB) of parity-protected capacity on the unit. The natural usage
that a drobo invites in users is to have a single, large device covering all the
data on device. For example, on Mac OS/X, users often create 16 TB LUNS on HFS.
This allows all the storage to fit on one large pool. The downside of larger
LUNS has to do with startup time, and the time to perform a file system check.</p>
<p>Under Linux unfortunately, with a first generation Drobo, one should limit the
volume size to 2 TiB[#gen12TiB]_. It is hoped, but not confirmed, that later
products support LUNS larger than 2 TiB on Linux. Drobom view therefore limits
lunsize to 2 TiB for the moment. The command line interface can be used to
create larger LUNS, they just might not work.</p>
<dl class="docutils">
<dt>ON LUNSIZES >= 2TB:</dt>
<dd><dl class="first last docutils">
<dt>-- On older distributions, there are a couple of gotchas related to</dt>
<dd><blockquote class="first">
<dl class="docutils">
<dt>linux tools which aren't 2TB ready... to exceed 2 TB, you need to:</dt>
<dd><dl class="first last docutils">
<dt>-- use GPT partitions, which aren´t supported by older fdisk</dt>
<dd>versions. Tools based on libparted work fine, mostly.</dd>
<dt>-- gparted fails, and seems to have a 1 TB limit on devices.</dt>
<dd>(bug #524948 reported to bugzilla.gnome.org) It's just the GUI,
as libparted is fine, and other tools based on it
still work.</dd>
</dl>
</dd>
</dl>
</blockquote>
<dl class="docutils">
<dt>-- on linux kernel < 2.6.24, the USB layer won't let one address</dt>
<dd>LUNs/offsets > 2 TB. For example, Ubuntu hardy (8.04) released in Spring
2008 has a 2.6.24, and so is OK. I've never been able to test this problem.</dd>
<dt>-- On linux kernel < 2.6.31 there is are reported firewire problem that will</dt>
<dd>prevent devices > 2 TiB from working.</dd>
<dt>-- ext3 with 4K blocks is supposed to allow file system capacity of 8 TiB.</dt>
<dd>4K blocks seem to be assigned by default. So I think a good max.
It would be fun to set the LunSIZE to 8 TiB and test it out...</dd>
</dl>
<p class="last">-- Windows XP does not support LUNS > 2 TiB</p>
</dd>
</dl>
</dd>
</dl>
<table class="docutils footnote" frame="void" id="gen12tib" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label">[7]</td><td>Many tests have been performed with first generation products
and several different failure modes have been found when exceeding 2 TiB.
<a class="reference external" href="http://www.drobo.com">Data Robotics</a> has addressed several failure modes, via fixes to the kernel
in 2.6.24, and for firewire in 2.6.31, and continues to address them in
later generation products.</td></tr>
</tbody>
</table>
</div>
<div class="section" id="what-kind-of-file-system">
<span id="filesystem-choice"></span><h2>What Kind of File System?</h2>
<p>Drobos work well under linux with ext3. One can, of course, set up an NTFS or
HFS+ or FAT32 if necessary, but it seems actively counter-intuitive on Linux.
Developers of Drobo-utils have not tested HFS. Linux does not write to Journalled HFS+
at this point, so HFS support is not present. Good success is reported with Ntfs-3g,
but the performance is much lower than what is typically reported with ext3.
Unless physical movement of the disk to between systems is required, the
native (ext3) format is the best option.</p>
</div>
<div class="section" id="id5">
<span id="iscsi-setup"></span><h2>iSCSI Setup</h2>
<p>This is a procedure for configuring a Drobo Pro for access via iSCSI. This
information is based on a post by Lemonizer on the Google Group 2009/05/16, with
updates based on improvements and tests by others in the fall of 2009:</p>
<pre class="literal-block">
1. Connect the Pro via USB, and manually configure the ip of the dbpro
# drobom info settings
# drobom set IPAddress 192.168.2.80
# drobom set NetMask 255.255.255.0
# drobom set UseStaticIPAddress True
</pre>
<p>The next step is to disconnect USB, and connect by iSCSI:</p>
<pre class="literal-block">
2. Install open-iscsi (http://www.open-iscsi.org/): sudo apt-get install open-iscsi
3. Connect the dbpro to host machine via iscsi
4. Query dbpro's id: sudo iscsiadm --mode discovery --type sendtargets --portal 192.168.2.80
5. Copy the id string returned by iscsiadm, something like "iqn.2005-06.com.datarobotics:drobopro.tdb091840080.node0"
6. Connect to the dbpro: sudo iscsiadm --mode node --targetname iqn.2005-06.com.datarobotics:drobopro.tdb091840080.node0 --portal 192.168.2.80:3260 --login
</pre>
<p>If everything went well, your drobopro should show up under /dev. Also check /var/log/messages to
confirm that the iscsi device connected successfully. After that, drobo-utils should be able
to detect the Drobo and manage it over ethernet/iSCSI.</p>
<p>(source: <a class="reference external" href="http://groups.google.com/group/drobo-talk/browse_frm/thread/453e02e105e9b41?hl=en">http://groups.google.com/group/drobo-talk/browse_frm/thread/453e02e105e9b41?hl=en</a> )</p>
<p>Some people reported data corruption. This link claims to fix one such
issue:
<a class="reference external" href="http://www.drobospace.com/forum/thread/13951/Dropped-iSCSI-connections/?page=2#24792">http://www.drobospace.com/forum/thread/13951/Dropped-iSCSI-connections/?page=2#24792</a></p>
</div>
</div>
<div class="section" id="drobo-firmware">
<h1>Drobo Firmware</h1>
<p>Upgrading firmware is pretty self-explanatory in the GUI. the first time you
press the Update button, it checks to see if a new firmware is available. If
it there is newer firmware, it offer to upgrade, with suitable prompts.
Similarly, the line mode interface has two commands to deal with firmware,
fwcheck will tell you if an upgrade is required. the fwupgrade
will do the job. It takes a few minutes, and prints a status
you you can see how it is progressing. Have patience:</p>
<pre class="literal-block">
root@pepino:/home/peter/drobo/drobo-utils/trunk# drobom fwupgrade
validateFirmware start...
Magic number validated. Good.
484 + 2937552 = 2938036 length validated. Good.
CRC from header: 4260378881, calculated using python zlib crc32: 398201869
yeah, the header CRCs do not match. For now they never do ... ignoring it.
CRC for body from header: 1852877921, calculated: 1852877921
32 bit Cyclic Redundancy Check correct. Good.
validateFirmware successful...
writeFirmware: i=484, start=484, last=2938036 fw length= 488
.
wrote 32768 bytes... total: 33252
wrote 32768 bytes... total: 66020
.
.
.
wrote 32768 bytes... total: 2720228
wrote 32768 bytes... total: 2752996
wrote 32768 bytes... total: 2785764
wrote 32768 bytes... total: 2818532
wrote 32768 bytes... total: 2851300
wrote 32768 bytes... total: 2884068
wrote 32768 bytes... total: 2916836
wrote 21200 bytes... total: 2938036
writeFirmware Done. i=2938036, len=2938036
root@pepino:/home/peter/drobo/drobo-utils/trunk#
</pre>
<p>when it's done, check if it worked using:</p>
<pre class="literal-block">
root@pepino# drobom status
/dev/sdf - 00% full - (['New firmware installed'], 0)
</pre>
<p>If the status is like that, then do:</p>
<pre class="literal-block">
root@pepino:/home/peter/drobo/drobo-utils/trunk# drobom shutdown
</pre>
<p>lights will flash etc... wait until Drobo goes dark.
Wait another five seconds, then un-plug the USB / connector.</p>
<p>Plug it back in, and wait 10 seconds.
it should start up with the latest firmware available for the drobo.</p>
<p>The drobom commands, like DRI's dashboard, will normally
get the latest and greatest firmware and upgrade. If necessary
one can load arbitrary firmware from the CLI with fwload command.</p>
<div class="section" id="safety">
<h2>SAFETY</h2>
<p>Those worried about safety of using this software should know: it was
developed with assistance from the vendor (<a class="reference external" href="http://www.drobo.com">Data Robotics</a> Inc.), and
in every case, based on vendor documentation, and with at least encouragement,
if not outright support. For each release, a QA.txt file is built, demonstrating the functionality tests run. There are multiple checksum verifications built
into the firmware upgrade process, so it is next to impossible to brick a drobo
using the tools. Drobo-utils verifies firmware checksums before attempting
to upload the image to the device, and the device checks the firmware against
the checksums as well. New firmware is loaded into an alternate location
from the currently active one, and if activation of the new firmware fails,
the drobo will simply boot the old one.</p>
<p>On the other hand, common sense rules do apply. Setting the LUN size, or
re-formatting a Drobo will erase all data whether it is done on Linux or
any other operating system. These are power tools, and they can do some
damage data if used without proper care. For example, the reliability
of any storage unit does not reduce the need for backups it only makes doing them
easier. A Drobo is an excellent place to put backups, but not a substitute for
them. Backups are the only way to address error 18 (the number of inches in
front of the keyboard the source of the issue lies.) and no storage unit can
protect against fire or flood.</p>
</div>
<div class="section" id="firmware-compatibility">
<h2>Firmware Compatibility</h2>
<p>Drobo has been tested with every old firmware version. Any Drobo should
be upgradable to modern firmware using the dashboard.</p>
<p>for Drobo v1's (only models available to me used for QA)</p>
<blockquote>
<dl class="docutils">
<dt>NOTE: really need at least 1.1.1 to use Linux & ext3.</dt>
<dd>just use the tools to upgrade firmware ASAP.</dd>
<dt>1.01 - very old... bad idea to install this, need to write</dt>
<dd>a script to get out, because it isn't in the revision
table. not much works except firmware upgrade.
DO NOT USE. UPGRADE ASAP</dd>
<dt>1.0.2 - works ok in CLI And GUI to view, and upgrade firmware.</dt>
<dd>DO NOT USE. UPGRADE ASAP</dd>
<dt>1.0.3 - GUI and CLI work OK, can upgrade firmware.</dt>
<dd>Used ntfs3g for a few months under ubuntu 7.10 Linux.
Used 2 TB LUN, with 1.5 TB of physical space available.
DO NOT USE. UPGRADE ASAP</dd>
<dt>1.1.0 - dashboard works no issues.</dt>
<dd><ul class="first last simple">
<li>from this point, you don't seem to need to unplug the USB
connector to complete the upgrade.</li>
<li>firmware prior to here deals badly with ext3.</li>
</ul>
</dd>
<dt>1.1.1 - 1.2.4 works without issues.</dt>
<dd>('name' not supported by firmware)</dd>
</dl>
<p>1.3.0 - works without issues.
1.3.5 - works without issues.</p>
</blockquote>
</div>
</div>
<div class="section" id="id6">
<span id="deployment"></span><h1>Deployment</h1>
<p>No storage unit ever constructed, at any price point, can live upto the expectation
of never losing data. There is no magic wand to wave to solve the data security
problem. People still need a strategy around backups and their maintenance.
Drobo makes it easier to implement a strategy, but does not replace it.
Data Robotics has a <a class="reference external" href="http://www.drobo.com/support/best_practices.php">best practices</a> page that says it well, but the phrasing
is a bit enterprisy, and while it provides general concepts, it is not
prescriptive enough for people to easily apply.</p>
<p>This section gives some examples & use cases to help people
develop the appropriate strategy for them. Try to keep it simple & concrete.</p>
<div class="section" id="general-concepts">
<h2>General Concepts</h2>
<p><em>Don't rely on a single device, ever!</em> Before deploying a storage unit,
one should perform the thought experiment of what will happen if all data on
it is lost. There are always levels of risk. For personal use, one might
accept the risk that if the house burns down, only have infrequent offsite
backups are available and months or years of data may be lost. If someone
is running their business out of their home, this risk will likely not be
acceptable.</p>
<p>The simplest method of backing up your data is to put it in a humungous
single place, and backup the whole thing. That is a valid strategy, but
consider the following realistic case:</p>
<p>A company does incremental backups[#incrbkup] once a day, and full backup
once a week. Once a month, a second full backup is kept as monthly, while the
weeklies are recycled. Monthly backups are kept for a year, and each
year, one backup is kept for five years. So if you write data once and
keep it unchanged for five years, you will have 3 weekly backups, 11 monthlies,
and 4 yearly backups of that data, or 17 copies. This strategy is not
unusual or particularly excessive, many corporate policies end up with 50
or more copies of the data.</p>
<p>With that in mind, if you avoid backing up what you don´t need to, then you
are not saving just one byte, but all the copies too. With a little thought,
one can usually reduce the total storage needs by classifying data appropriately.</p>
<table class="docutils footnote" frame="void" id="incrbkup" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label">[8]</td><td>An incremental backup is where only what has changed since the last full backup is saved. a full backup is a complete copy of all data.</td></tr>
</tbody>
</table>
<p>Different Drobo models have very different performance. Deploying a Drobo
using a USB interface as primary storage is likely to disappoint.
See the compatibility matrix for details.</p>
<p><em>Classify Your Data</em></p>
<p>There are different levels of value for data. Things that are downloaded
from the internet, or source code that is pushed to a repository on the internet,
have natural backups in most cases. The loss of data being queued for printing,
might not be a terrible loss. The loss of videos recorded off the air from
television, might not be terrible either.</p>
<p>Any kind of data which is either not worth backing up, or for which a backups
already exist, does not need to be backed up locally. The other end of the
spectrum is proprietary data, for which copies on the internet are not be
appropriate, and which is irreplaceable if lost. In a photographic business,
the photos, Tax records, accounts, etc... In a personal realm, these would
include family photos, etc... lets call this sort of information <em>precious data</em></p>
<p>That irreplaceable data is what you need to safeguard. So the classification
can be done in a number of ways, but the simplest is just to only put precious
data in the home directory. So far that´s normal. The unusual thing comes next: Do not put anything else there. Internet downloads, easily replicable data, etc...
should go somewhere else. In general, keep the home directories of users as small
and precious as possible.</p>
<p>As another example, in my personal use case, Linux is readily downloadable, so
there are no system backups at all. Configurations are relatively straightforward,
only credentials, and special configurations are backed up, by having copies
of the information in an normal /home directory of a user. The restoral time
for a single system is not an issue for my personal use, and by the time a
restore is necessary, there will probably a new OS version to try out,
so the value of system backups is quite limited.</p>
<p>The only thing backed up, is the personal (/home) directories of a few users.</p>
</div>
<div class="section" id="primary-storage">
<h2>Primary Storage</h2>
<p>If a Drobo is used for primary storage for precious data, a second one should be
obtained as a backup device. It´s as simple as that. Keeping all data on
one device that cannot be backed up is asking for trouble.</p>
<p>Zealots will say that the second unit should be off-site.
The Gen 1 / Gen 2 Drobo´s are also have limited performance, and are perhaps not
well suited to a role as primary storage.</p>
</div>
<div class="section" id="scratch-storage">
<h2>Scratch Storage</h2>
<p>If the data there is all space that either exists elsewhere (mirrors of internet
sites), can be regenerated (object files of compilation, recordings from on-air
broadcasts in a media server), none of this data is particularly precious, and
all of it can be recovered over time in the event of a data loss on one unit.</p>
<p>Again, one argument against such usage is performance. First and second generation
units are a bit on the slow side for use in say. On the other hand, media
serving is a streaming application with typically low instantaneous bandwidth
requirements, so even first generation Drobos should be fine for that.</p>
</div>
<div class="section" id="backup-repository">
<h2>Backup Repository</h2>
<p>A rational means of configuring the Drobo is as a backup repository.
The drobo is destination of the backups. Primary copies are on the
desktops & laptops being backed up. Viewed in that way, backup data
is easily recovered in the event of a data loss: just backup the system
again. Of course the history of backups is lost, but the important
thing is usually being able to recover current data.</p>
</div>
</div>
<div class="section" id="troubleshooting">
<h1>Troubleshooting</h1>
<div class="section" id="no-drobos-discovered">
<h2>No Drobos Discovered</h2>
<p>To find Drobo on a system, drobo-utils queries all the attached devices for indications
it is made by <a class="reference external" href="http://www.drobo.com">Data Robotics</a>. These strings change from product to product.
If the (new model) Drobo is not detected, then run the command line interface
with the hardware detection debugging output turned out. like so:</p>
<pre class="literal-block">
# drobom -v 16 status
examining: /dev/sda
id: (0, 0, 0, 0, 'ATA ')
rejected: vendor is ATA (not from DRI)
examining: /dev/sdb
id: (2, 0, 0, 0, 'ATA ')
rejected: vendor is ATA (not from DRI)
examining: /dev/sdc
id: (8, 0, 0, 0, 'Drobo ')
rejected: vendor is Drobo (not from DRI)
returning list: []
No Drobos discovered
</pre>
<p>Here you see that the vendor string is 'Drobo' which was not a known vendor string
at the time this example was run. so then try:</p>
<pre class="literal-block">
# drobom -s Drobo status
</pre>
<p>In other words, take the unknown vendor string and feed it as -s option to tweak
detection of drobom. Your drobo will likely then be picked up.</p>
</div>
<div class="section" id="only-one-lun-shows-up">
<h2>Only One LUN Shows up?</h2>
<p>LUN is an abbreviation of 'Logical UNit'. The origin of the term is
SCSI[#SCSI]_ terminology.
When RAID units became too large for support in the past, and were sub-divided
to present smaller units the operating system. The default LUNSIZE on Drobos
is 2 TiB (adjustable using the tools.) If more disk space (after allowing for
parity/redundancy) than LUNSIZE is installed in a unit, Drobo will show a
second (or even third) LUN. Each LUN shows up in Linux as a separate disk
(examples if the first LUN shows up as /dev/sde, the next will be /dev/sdf,
then /dev/sdg.)</p>
<p>If you think you should see multiple LUNS and you don't, you might have a look at
some kernel settings: make sure that scsi_mod kernel module is loaded, make
sure /sys/module/scsi_mod/parameters/max_luns is > 1.</p>
<p>Drobom view will start up one GUI per drobo, regardless of the number
of LUNS. If asked to format, all LUNS for the device will be formatted.</p>
<table class="docutils footnote" frame="void" id="scsi" rules="none">
<colgroup><col class="label" /><col /></colgroup>
<tbody valign="top">
<tr><td class="label">[9]</td><td>Small Computer System Interface. A ubiquitous standard for computers to
communicate with storage hardware. SCSI includes hardware cabling specifications,
which are mostly obsolete, but what remains is the "command set", the language used
by the computer to make requests to the device. In that sense, All Drobos are SCSI devices.
The SCSI commands are tunnelled within other protocols used to transport data between
computer and device (Firewire, USB, eSATA, and, yes... ISCSI)</td></tr>
</tbody>
</table>
</div>
<div class="section" id="getting-help-from-dri">
<h2>Getting Help from DRI</h2>
<p>DRI intends Drobospace.com for owners to talk with one another, except no non-owner
can see the discussions, and early on, there was a lot of input from DRI staff, so
it looked a lot like a support forum, but it really isn't. A lot of owners
objected to these forums being private, so a google group was started for people
to talk with one another, and the discussions to remain public:</p>
<p><a class="reference external" href="http://groups.google.com/group/drobo-talk/topics?hl=en">http://groups.google.com/group/drobo-talk/topics?hl=en</a></p>
<p>There is still a great role for Drobo space, in that Tier 3 support analysts
(essentially developers.) sometimes look over there. For tier 3 support, one
cannot expect guaranteed response time, but one may be able to provide some
input into future products or firmware features. It turns out that the
Drobospace forums aren't really for support. but don't take my word for it,
here is above was DRI's take (verbatim from a post on
drobospace by MarkF 2008/08/29) on things:</p>
<p>To contact <a class="reference external" href="http://www.drobo.com">Data Robotics</a> Inc. for support your options are:</p>
<ol class="arabic simple">
<li>phone support - technical issues: 1-866-426-4280, Mon-Fri from 8am-5pm PST, excluding Holidays.</li>
<li>phone support - presales questions: 1-866-99ROBOT</li>
<li>email support - technical issues: <a class="reference external" href="mailto:support@drobo.com">support@drobo.com</a></li>
<li>email support - presales questions: <a class="reference external" href="mailto:sales@drobo.com">sales@drobo.com</a></li>
<li>web-based support request: <a class="reference external" href="http://www.drobo.com/Support/Request_Support.html">http://www.drobo.com/Support/Request_Support.html</a></li>
</ol>
<p>All technical support calls, emails, and web requests are assigned a case number
and tracked. DRI has 3 tiers of customer support. Tier 1 and 2 handle the majority
of cases. They are responsible for tracking phone, email, an web cases and
resolving them. Some cases are escalated to Tier 3 whose personnel reside
in our corporate headquarters and have access to engineering staff.</p>
<p>Support on Drobospace.com:</p>
<p>Drobospace.com is a user community and relies on the volunteer efforts of its members to help each other. Because it is run by volunteers response to problems varies. Tier 3 support personel monitor the Drobospace forums -- tiers 1 and 2 are focused directly on customers, and they are not required to read drobospace.com. By design Tier 3 personnel do not immediately respond to each posting in order to allow the community to function. Depending on the nature of the problem, tier 3 may post in the forum or contact the member directly through a private message to facilitate problem resolution.</p>
<p>Because Drobospace is owned and run by a third party, Capable Networks LLC, there is no linkage with DRI's database systems. Posts here are not assigned case numbers and tracked - that only happens with cases opened directly with DRI.</p>
</div>
</div>
<hr class="docutils" />
<div class="section" id="faq">
<h1>FAQ</h1>
<div class="section" id="what-lunsize-should-i-use">
<h2>What LUNSIZE should I use?</h2>
<p>2 Terabytes is the biggest you should use for now. There are lots
of experiments on the google groups, summarized here: <a class="reference internal" href="#lunsize-fits-all">LUNSIZE fits all</a>
Also consult the compatibility matrix indicates best guesses at the
current state of affairs. DRI announced new firmware 1.1.4 for Drobo PRO
which is supposed to remove the 2 TiB restriction, but that isn't confirmed
yet.</p>
</div>
<div class="section" id="what-happens-when-i-used-a-bigger-lunsize">
<h2>What Happens When I Used A Bigger LUNSIZE?</h2>
<p>That's actually a bit nasty. Nothing happens at first, everything seems
to work fine. After a while, it fails to reclaim space when files are deleted.
The blue capacity lights don't show much relation to how full the file system
is, as reported by the operating system. Drobo may become insatiable, always
asking for more and more disk space, even though the amount of data used
on the file system doesn't warrant it. In extreme cases, data may become
in-accessible.</p>
</div>
<div class="section" id="how-do-i-check-that-a-lunsize-works">
<h2>How Do I Check That a LUNSIZE Works?</h2>
<p>DRI naturally releases new versions of firmware and may fix these issues
at some point. If you are willing to test it out on your new Drobo, the
procedure to do so is simple:</p>
<pre class="literal-block">
1. Create a file system as per normal.
2. Fill the physical space up. (blue lights should light up.)
3. Remove alot of the data.
</pre>
<p>If the problem is not there, then blue lights will function properly and go
out to correspond to the deleted data after a while. If the blue lights
do not go out after step 3, then do not trust your data to this file system.
Re-create with a smaller LUN, and try again. 2 Terabytes is the only case
of documented success so far.</p>
</div>
<div class="section" id="can-i-user-reiserfs-xfs-btrfs-xxfs">
<h2>Can I user ReiserFS, XFS, BTRFS, xxfs ?</h2>
<p>Short answer: no.</p>
<p>For Drobo to perform storage management, it has to know what space is free, so it
needs to understand the file system you are using. The list of file systems
it official understands is: FAT32, NTFS, HFS+, ext3. That's it, so if you want
to use reiserfs, or xfs, or GFS, or whatever... you are doing research. The
vendors says those other file systems types will not work.</p>
</div>
<div class="section" id="if-my-drobo-breaks-can-i-get-my-data-back">
<h2>If my Drobo breaks, Can I Get My Data back?</h2>
<p>No. The way the data is placed on the disks is completely proprietary.
You cannot take the drives and connect them individually to a server, and read
the data off that way, because it isn't a linux md or lvm format that can easily
be reconstructed. You cannot give the disk pack to a data recovery company,
because they do not know the data format either, and you will have to pay them
to reverse engineer DRI's format, which will get expensive quickly.</p>
<p>You need either a backup, or another Drobo. Even among Drobos there
are limits to compatibility see the Drobo.com web site for details.</p>
</div>
<div class="section" id="my-usb-drobo-always-comes-up-as-a-different-disk">
<h2>My USB Drobo always comes up as a different Disk!</h2>
<p>The order and timing of disks being connected to hot-plug busses will
determine the device name (it might be /dev/sdb one time, and /dev/sdc another.)
So putting /dev/sdX in the fstab to mount their disks, as is traditionally done,
won't work. Instead, do:</p>
<pre class="literal-block">
peter@pepino:~$ ls -l /dev/disk/by-uuid
total 0
lrwxrwxrwx 1 root root 10 2009-12-15 04:54 2C88743C8874071C -> ../../sda3
lrwxrwxrwx 1 root root 10 2009-12-15 04:54 32a41d0a-b193-41f3-86fa-29bbee8cd2b3 -> ../../sda8
lrwxrwxrwx 1 root root 10 2009-12-26 12:08 3cd5d9cc-c227-4ed8-bab2-60c2d71f6e9d -> ../../sdf1
lrwxrwxrwx 1 root root 10 2009-12-15 04:54 72b0ee8c-d0e8-479d-b79c-3dbda1581f55 -> ../../sda6
lrwxrwxrwx 1 root root 10 2009-12-15 04:54 814472db-dbee-411c-8870-7ca59f32e7c1 -> ../../sda5
lrwxrwxrwx 1 root root 10 2009-12-26 12:16 8ed93296-9be2-4576-9ae4-9d9c78363fb6 -> ../../sdg1
lrwxrwxrwx 1 root root 10 2009-12-15 04:54 a4bc252e-0eb7-489c-94e7-688efd528665 -> ../../sda7
lrwxrwxrwx 1 root root 10 2009-12-15 04:54 bc1ab400-df49-457d-8700-c77dde19e450 -> ../../sda2
lrwxrwxrwx 1 root root 10 2009-12-15 04:54 C2EE700DEE6FF7D5 -> ../../sda1
peter@pepino:~$
</pre>
<p>The UUID is a name that is constant for a partition. Each time a partition
is mounted, a link will be created in this directory towards the correct
/dev/sdX. A UUID related /etc/fstab entry looks like:</p>
<pre class="literal-block">
UUID=3cd5d9cc-c227-4ed8-bab2-60c2d71f6e9d /drobo01 ext3 defaults 0 2
</pre>
</div>
<div class="section" id="how-come-the-command-to-build-a-file-system-builds-an-ext2-file-system">
<h2>How come the command to build a file system builds an ext2 file system?</h2>
<p>because an ext3 file systems is an ext2 file system with a journal.
The normal command to build an ext3 file system is mke2fs -j.</p>
</div>
<div class="section" id="can-you-have-different-luns-with-different-file-systems-on-them">
<h2>Can you have different LUNS with different file systems on them?</h2>
<p>DRI: Multiple partitions per LUN is supported. Having any combination of file
supported file systems on the different LUNS and partitions is fine as well.</p>
</div>
<div class="section" id="does-drobo-work-with-lvm">
<h2>Does Drobo work with LVM?</h2>
<p><a class="reference external" href="http://www.norio.be/blog/2008/11/setting-drobo-linux">Some people do it</a>. I would not risk it.</p>
<p>The Linux Volume Manager is a layer of software which is shimmed between the file system layer, and the physical disks. It provides a 'fake' (virtual) volume on which file systems are built. This gives flexibility to concatenate several physical volumes together to make a single file system, or allocate a single volume to different file systems over time, as needs dictate rather than all at the outset.</p>
<p>For Drobo, LVM would be especially cool in that one could initially allocate only
the physical space actually available within the LUN, and thus applications which
key on avoiding filling file systems would function correctly, instead of always
asking to insert more storage, and not managing the storage available. When more
physical space (new drives!) becomes available, one could allocate more space to
the virtual volume, and then grow the file system. So Drobo would still take
care of the drudgery of RAID set maintenance, relay operations, etc... but the
user would have more control on how space was allocated at the OS level.</p>
<p>For that to work, Drobo firmware would have to understand LVM to some extent.
Initial experiments showed that the firmware didn't appreciate LVM. Being
able to add a Drobo unit as an pv to an LVM unit would be very cool.</p>
</div>
<div class="section" id="does-drobo-support-full-disk-encryption">
<h2>Does Drobo Support Full Disk Encryption?</h2>
<p>Drobo works by knowing how the file system is laid out and pretending to have
more space than is physically present. Drobo does some sleight of hand to
manage disk space and move things around optimally as hard disks fail or are
added.</p>
<p>Full disk encryption implemented by the operating system makes the Drobo unable
to understand the file system, so it doesn't know which blocks are in use, and
the unit will always believe the system is completely full. Drobo will not
behave well. Among the methods that will not work are any that operate on a
raw disk partition, such as truecrypt, or any of the linux cryptoloop based solutions.</p>
<p>Instead of whole disk encryption, a method that uses an underlying file system
that is well known to the Drobo (the list is short: FAT32, NTFS, HFS, EXT2) is
needed. On windows, encrypting directories with standard NTFS will work fine.
On Linux, a good choice would be EncFS <a class="reference external" href="http://www.arg0.net/encfs">http://www.arg0.net/encfs</a>, which encrypts
file names and data over an ext file system, or some other method which uses
FUSE <a class="reference external" href="http://fuse.sourceforge.net">http://fuse.sourceforge.net</a>. is reported to work well.</p>
</div>
<div class="section" id="i-have-read-everything-help">
<h2>I have read everything. Help?</h2>
<p>Best first stop is the google <a href="#id9"><span class="problematic" id="id10">Group_</span></a>.</p>
</div>
<div class="section" id="what-is-drobo-utils-license">
<h2>What is Drobo-utils' License?</h2>
<p>General Public License - Gnu - GPL .</p>
</div>
</div>
<div class="section" id="credits">
<h1>Credits</h1>
<p>who did what:</p>
<pre class="literal-block">
Peter Silva: wrote most all of it.
Chris Atlee: the proper debian packaging.
Brad Guillory: some help with diagnostics and patches.
Joe Krahn: lots of inspiration.
Andy Grover: some elegance cleanups.
Sebastian Sobolewski: DroboPro patches, and testing.
</pre>
<p>Testers (of DroboPRO):
robj, Sebastian (aka Tom Green), ElliotA, Andrew Chalaturnyk</p>
</div>
<div class="section" id="administrivia">
<h1>Administrivia</h1>
<p>version 9999, somewhen</p>
<p>copyright:</p>
<p>Drobo Utils Copyright (C) 2008,2009 Peter Silva (<a class="reference external" href="mailto:Peter.A.Silva@gmail.com">Peter.A.Silva@gmail.com</a>)
Drobo Utils comes with ABSOLUTELY NO WARRANTY; For details type see the file
named COPYING in the root of the source directory tree.</p>
</div>
<div class="system-messages section">
<h1>Docutils System Messages</h1>
<div class="system-message" id="id7">
<p class="system-message-title">System Message: ERROR/3 (<tt class="docutils">README.txt</tt>, line 62); <em><a href="#id8">backlink</a></em></p>
Unknown target name: "darfs".</div>
<div class="system-message" id="id9">
<p class="system-message-title">System Message: ERROR/3 (<tt class="docutils">README.txt</tt>, line 981); <em><a href="#id10">backlink</a></em></p>
Unknown target name: "group".</div>
</div>
</div>
</body>
</html>
