Drobo 5N Technical Details

There are a few changes I’ve noticed while poking around the system:
[list][*]It appears we have a quad-core Marvell CPU (3 cores reported; I assume like the FS one core is dedicated to BeyondRAID):

[code]$ cat /proc/cpuinfo
Processor : Marvell PJ4Bv7 Processor rev 2 (v7l)
processor : 0
BogoMIPS : 1196.85

processor : 1
BogoMIPS : 1196.85

processor : 2
BogoMIPS : 1196.85

Features : swp half thumb fastmult vfp edsp vfpv3 tls
CPU implementer : 0x56
CPU architecture: 7
CPU variant : 0x2
CPU part : 0x584
CPU revision : 2

Hardware : Marvell Armada XP Development Board
Revision : 0000
Serial : 0000000000000000[/code]
[*]We have a LOT more memory to play with. Nearly 1GB available to Linux:

$ cat /proc/meminfo | head -1 MemTotal: 880712 kB
[*]Linux kernel is now 3.2; remains to be seen just how said kernel has been configured.

$ uname -a Linux Drobo 3.2.27 #1 SMP Tue Jan 15 13:57:54 PST 2013 armv7l GNU/Linux
[*]glibc is now updated to 2.11.1

$ /lib/libc.so.6 GNU C Library (EGLIBC) stable release version 2.11.1, by Roland McGrath et al. Compiled by GNU CC version 4.4.5 20100614 (prerelease). Compiled on a Linux >>2.6.32-28-server<< system on 2011-06-08. Available extensions: crypt add-on version 2.1 by Michael Glad and others GNU Libidn by Simon Josefsson Native POSIX Threads Library by Ulrich Drepper et al Support for some architectures added on, not maintained in glibc core. BIND-8.2.3-T5B
[*]Still using busybox, but updated to 1.20.2.

BusyBox v1.20.2 (2012-10-05 10:58:47 PDT) multi-call binary.

[*]My DroboFS diskpack is mounting as ext4; this is a bit odd since the FS itself used ext3, and I wasn’t aware of any plans to perform an in-place filesystem upgrade. EDIT: According to ricardo, this is likely just backwards-compatible mounting.

$ mount <snip> rootfs on / type rootfs (rw) /dev/root on / type jffs2 (rw,relatime) proc on /proc type proc (rw,relatime) devpts on /dev/pts type devpts (rw,relatime,gid=5,mode=620) tmpfs on /tmp type tmpfs (rw,relatime) none on /sys type sysfs (rw,relatime) /dev/shm on /dev/shm type tmpfs (rw,relatime,size=51200k) /dev/mtdblock1 on /var type jffs2 (rw,relatime) /dev/sda1 on /mnt/DroboFS type ext4 (rw,noatime,nodiratime,barrier=1,data=ordered,discard)
[*]No change to the size of the root partition; perhaps more shared memory space:

$ df -h Filesystem Size Used Available Use% Mounted on rootfs 30.0M 26.9M 3.1M 90% / /dev/root 30.0M 26.9M 3.1M 90% / tmpfs 430.0M 144.0K 429.9M 0% /tmp /dev/shm 50.0M 256.0K 49.8M 1% /dev/shm /dev/mtdblock1 3.0M 500.0K 2.5M 16% /var /dev/sda1 16.0T 4.5T 11.4T 28% /mnt/DroboFS
[*]The diskpack is still mounted at “/mnt/DroboFS”, but “/mnt/DroboNAS” is symlinked to it. There’s a mountpoint for “/mnt/Drobo5N”, which I assume is used for native, non-migrated diskpacks. Nice that we can base everything around a single path, however.

$ ls -l /mnt/ drwxr-xr-x 2 root root 0 Jan 15 13:59 Drobo5N drwxr-xr-x 7 root root 4096 Jan 18 17:20 DroboFS lrwxrwxrwx 1 root root 7 Jan 15 13:59 DroboNAS -> DroboFS
[*]They’re definitely laying the groundwork for permissions changes. Everything I’ve seen so far is mapped to user/group of “default:default”. I had to re-chown my home folder before pubkey authentication would work again.[/list]

As I find out more, I’ll post back.

Man, I completely missed this new forum. Anyway, let’s dissect this data:

That confirms what Jeff Long told me on the phone.

After a bit of googling, I found the Marvell Armada XP board product page 1. More specifically, the 5N seems to be based on the MV78460 (quad-core version). According to that site, this CPU should have a 2MB cache, which is probably split the same way as the FS’s (512KB per core). But more importantly, this CPu is rated up to 1.6 GHz. If I read your ‘cat /proc/cpuinfo’ correctly, the 5N has them at 1.2 GHz. I wonder if you can overclock the 5N… :slight_smile:

Other nice things to point out: this CPU is an ARMv7, which means it is two generations younger than the FS’s (ARMv5). In particular, this means that it has a good floating point unit (see the “vfpv3” feature). I wonder if this means that the 5N will be able to transcode video on the fly – I’m thinking Plex Media Server, or even Mediatomb.

Hallelujah!

Man, I’m so giddy right now. This opens so many doors! The aforementioned Plex Media Server might actually run perfectly on this thing. No to mention all of the other apps at the same time! It might be too soon to say goodbye to high loads, but I feel very optimistic about it.

Fantastic. This is a recent enough kernel, so pretty much all of the new goodies should be there. Also, this probably means that the hardware crypto engine (to accelerate AES and SHA1) should be well supported. I expect great things for OpenSSH on this platform.

I’m no ext expert, but as far as I remembered the ext family is backwards compatible 2. Mounting an ext3 partition as ext4 should have no negative side-effect. Mental note for the future: investigate if it is possible to convert an ext3 pack to ext4. In theory it should be possible.

Well, it seems that they have been listening to us after all. One can only hope that they will introduce the same level of support for DroboApps, i.e., make it easy to create users/groups to quarantine apps that are exposed to the internet.

If you would care to answer a few more questions, here’s what I would like to know:
[list]
[]What about the flash memory? Do we get more of it? How much free space is in there?
[
]Is the user environment still based on busybox? What version do you get when you run “busybox --help | head -n 1”? (For reference, the FS returns “BusyBox v1.14.2 (2009-07-29 17:47:47 PDT) multi-call binary”)
[]What is the output when you run “/lib/libc.so.6”? The FS is uses libc 2.5, which is pretty old by now. In that same output, which was the compiler used? The FS was compiled with GCC 4.2.0, which again, is very much deprecated nowadays.
[
]What about the swap area? Is it still on a file on the disk pack? I assume that on your legacy diskpack it will still be on a file, but I wonder if in new packs the area will be allocated on a separate partition.
[/list]

I like the fact that they chose to stay with an ARM processor. This means what you already experienced: most DroboApps should just run on the 5N. The complication, of course, is that now you can have an app deployed on a 5N and it can be installed in two distinct paths: /mnt/DroboFS or /mnt/Drobo5N. That might make things considerably harder for some apps.

Running all of my apps (including stgt, crashplan, etc) shows the Drobo 5N at 100% idle - sometimes it goes as low as 99.9%. Now that’s with nothing going on - no shares mounted, no backups running, etc - but even under such conditions on the FS we’d see a few percent used. Not so here. Load averages are around 0.65.

[quote=“diamondsw, post:1, topic:41188”]
I wasn’t aware of any plans to perform an in-place filesystem upgrade. Perhaps that’s also what’s churning away. … This also may explain the ongoing disk activity I’ve seen on my 5N since I put in my FS diskpack - chown’ing a few million files will take a while.[/quote]

(FYI, I don’t think either of these were ultimately the case. While it mounts as ext4 and the permissions on the root partition are different, it doesn’t appear that the disk pack was actually changed. Good from a data stability standpoint, and we can always chown as needed.)

[quote=“ricardo, post:2, topic:41188”]
If you would care to answer a few more questions, here’s what I would like to know:
[list]
[*]What about the flash memory? Do we get more of it? How much free space is in there?[/quote]

Doesn’t appear that’s changed - but I don’t recall having 50MB of shared memory. Might be a nice playground for temp files and the like (I redirected my crashplan service.log there).

$ df -h Filesystem Size Used Available Use% Mounted on rootfs 30.0M 26.9M 3.1M 90% / /dev/root 30.0M 26.9M 3.1M 90% / tmpfs 430.0M 144.0K 429.9M 0% /tmp /dev/shm 50.0M 256.0K 49.8M 1% /dev/shm /dev/mtdblock1 3.0M 500.0K 2.5M 16% /var /dev/sda1 16.0T 4.5T 11.4T 28% /mnt/DroboFS

[quote=“ricardo, post:2, topic:41188”]
[*]Is the user environment still based on busybox? What version do you get when you run “busybox --help | head -n 1”? (For reference, the FS returns “BusyBox v1.14.2 (2009-07-29 17:47:47 PDT) multi-call binary”)[/quote]

Still busybox:

BusyBox v1.20.2 (2012-10-05 10:58:47 PDT) multi-call binary.

[quote=“ricardo, post:2, topic:41188”]
[*]What is the output when you run “/lib/libc.so.6”? The FS is uses libc 2.5, which is pretty old by now. In that same output, which was the compiler used? The FS was compiled with GCC 4.2.0, which again, is very much deprecated nowadays.[/quote]

$ /lib/libc.so.6 GNU C Library (EGLIBC) stable release version 2.11.1, by Roland McGrath et al. Compiled by GNU CC version 4.4.5 20100614 (prerelease). Compiled on a Linux >>2.6.32-28-server<< system on 2011-06-08. Available extensions: crypt add-on version 2.1 by Michael Glad and others GNU Libidn by Simon Josefsson Native POSIX Threads Library by Ulrich Drepper et al Support for some architectures added on, not maintained in glibc core. BIND-8.2.3-T5B

[quote=“ricardo, post:2, topic:41188”]
[*]What about the swap area? Is it still on a file on the disk pack? I assume that on your legacy diskpack it will still be on a file, but I wonder if in new packs the area will be allocated on a separate partition.
[/list][/quote]

Can’t say for a new diskpack (don’t think anyone will be able to for a while), but I still have the same swapfile in use.

[quote=“ricardo, post:2, topic:41188”]
The complication, of course, is that now you can have an app deployed on a 5N and it can be installed in two distinct paths: /mnt/DroboFS or /mnt/Drobo5N. That might make things considerably harder for some apps.[/quote]

I’m still wondering about how the “DroboNAS” mountpoint is being used - it’s symlinked to FS on mine; perhaps it’s symlinked to 5N on a new diskpack. Then we could just target 5N apps at “DroboNAS” - although it would be mildly annoying to support both 5N and FS apps.

cool diamond, you got your 5n :slight_smile:

TL;DR: Excellent news all around. Let’s break it down below:

I would imagine it so. With 3 cores, each one at least 40% faster than the FS’s this thing should fly. I wonder what is the output of “openssl speed -evp aes-128-cbc” is like. This is what the FS prints out:

# openssl speed -evp aes-128-cbc
Doing aes-128-cbc for 3s on 16 size blocks: 1062722 aes-128-cbc's in 2.90s
Doing aes-128-cbc for 3s on 64 size blocks: 307983 aes-128-cbc's in 2.93s
Doing aes-128-cbc for 3s on 256 size blocks: 77213 aes-128-cbc's in 2.84s
Doing aes-128-cbc for 3s on 1024 size blocks: 19897 aes-128-cbc's in 2.89s
Doing aes-128-cbc for 3s on 8192 size blocks: 2480 aes-128-cbc's in 2.87s
OpenSSL 1.0.1c 10 May 2012
built on: Thu Aug 16 16:20:43 CEST 2012
options:bn(64,32) rc4(ptr,char) des(idx,cisc,16,long) aes(partial) idea(int) blowfish(ptr) 
compiler: /usr/local/arm-2007q1/bin/arm-none-linux-gnueabi-gcc -DOPENSSL_THREADS -D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H -DL_ENDIAN -DTERMIO -march=armv5te -mtune=arm926ej-s -Os -mthumb -ffunction-sections -fdata-sections -fomit-frame-pointer -Wall
The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
aes-128-cbc       5863.29k     6727.27k     6960.05k     7050.01k     7078.80k

I think we did get a bit more. This is my output (keep in mind that I have added some files to the flash area, so free space is a bit less that it should:

# df -h
Filesystem                Size      Used Available Use% Mounted on
/dev/root                21.0M     19.7M      1.3M  94% /
/dev/shm                 30.0M         0     30.0M   0% /dev/shm
/dev/mtdblock/2           2.0M    612.0K      1.4M  30% /var
/dev/sda1                16.0T      4.6T     11.4T  29% /mnt/DroboFS

The difference in /dev/shm is a matter of configuration. They just decided to allocate a larger chunk or RAM, which I think it’s fine, given the large amount available now.

There is something different about the layout of the flash chip, though. Notice that /var is now mounted in /dev/mtdblock1, instead of /dev/mtdblock/2. They probably shuffled the flash “partitions” around for whatever reason.

But a much more recent one (Oct '12 – damn, that thing was compiled on the last day before shipping or something). I’m not very familiar with busybox, but I hope that now the commands are closer to the their full-fledged counterparts.

Excellent. A much more recent libc and a pretty recent GCC. This should make much easier to port some of the more recent software (such as rtorrent, which requires some features of GCC 4.3).

Yeah, I’m thinking about apps that need full paths on their configuration files. This means that pretty much all of those apps are going to need an install script that will replace the path as needed.

www.involvinglouisvuitton.com USDA may be the case overestimating the amount having to do with corn that will be dealt with into ethanol everywhere over the 2008/09com Realty has reports covering topics some of these as it and will probably be happy for additional details on furnish upon request

ugg outlet online These are do nothing more than an all in one couple having to do with innovations as part of your genre about movie distribution so as to provide going to be the converging TV and Internet industries The bad news is because that element usually hard,and can be painful

ugg boots outlet First, addiction to learn more about prescription drugs at least alcohol will be the an equal move to make murderer that affects all of them are races,people back and forth from they all are occupations, religions, nationalities, socio-economic groups and back and forth from most of them are political parties But if you’d like for more information on synchronize going to be the calendar allowing an individual going to be the sun Caesar had to let going to be the a history of year drag all around the along with 445 days

ugg outlet How Can You Protect Yourself from ScammersMany people which of you purchase going to be the mouse click bank data entry service or product fail and give up so that you have data entry little tasks 5

uggs outlet Melfi should be a great deal more regarding an all in one challenge as well as her Pakistan is this all over the trouble providing some one going to be the United States and they are renegotiating his or her pact

ugg outlet You can’t change going to be the past or at least forecast going to be the serious,and for that reason be the case articles or blog posts providing some one what you have today as well as What academic programmes are run because university?or Our academic disciplines in are varied

ugg boots outlet Who is the reason that allowing you to have my hand?Not further ago my wife and I happen to have been planning a multi functional get-together leaving a few of the acquaintances and wanted to learn more about support a cheese board after dinnerThe primary measure to do with a short while ago ratings,which determine going to be the money having to do with that a period of time being bought and can be bought,usually AC Nielsen, an independent company all of which can make it information as to learn more about which of you watches what all over the television

ugg outlet In this sense almost any common stock,for the as a resource box is this everywhere over the online earnings, entails a few of the risk,among the speculation Label printing acts as an all in one great safety online as well as for situations like this

uggs outlet It begins all around the that date one of the more and then for cultures that use a 365-day solar calendar Cast and characters See also: List having to do with characters back and forth from The Sopranos The Sopranos features a number of different characters all through its six-season run

ugg outlet The first reason a good reason online dating is this more preferable than real part of the world dating is this that of anonymity When dealing allowing you to have strong personalities,let them are aware of that everywhere in the one or more assertive manner what may not be acceptable to you