DSM-G600, DNS-3xx and NSA-220 Hack Forum

First - think about what would be the end result if you, in good faith, followed the instructions provided by myself or any one else and things went awry.  Would you blame us or would you accept responsibility for your own data.

RAID1 is not a backup - if you didn't understand that before, I'm sure you understand it now - what does a 750GB drive cost now?  Is it more than your data is worth to you?  Sure I recognize that transferring 600GB of data across the network takes time, but would it take less time to recreate it?

It's your data, it's your time, it's your money - it's your choice - and if you haven't backed it up - it's your risk.

Now to deal with your question - removing the disks from one DNS-323 and installing them in a second should not cause or require a reformat - although depending on the last used configuration of the second unit, what exactly happens next may differ.

If it is a brand new unit in a "factory default" configuration (standard volumes), you should simply end up with two standard volumes, each containing a copy of your data.

If the last used configuration was RAID1 - you should be able to swap the drives from the first unit to the second without a hitch (although I don't guarantee it will be without a glitch which is why you should back up the data before doing anything)

If the last used configuration was JBOD or RAID0 - I have no idea what you should expect, not having done any tests with these configurations.

So - based on the above - I'd say the easiest way is to add a pair of drives (any size) to your replacement DNS-323, configure it for RAID1 and then swap the 750s in - whether or not you backup is, of course, your choice.

Last edited by fordem (2008-05-09 15:06:56)

jdoering

What I have done is swapped back & forth between two pairs of RAID1 disks - a pair of 80GB Maxtors and a pair of 250 GB Seagates - it doesn't seem to mind which pair is installed.

I lack the linux knowledge to go poking around the innards of the unit - as well as to some extent the curiosity - over the years I never had the need to know how and Adaptec or AMI (now LSI Logic) RAID controller worked, just how to manipulate it through the management utilities, and I don't see why I should treat this unit any different.

From that viewpoint a good backup is all I need - I can swap controllers, I can swap disks - I can get the end users' system backup and running and that essentially is what matters.

One thing that might have been nice - I don't know if it's possible given the fact that this is software RAID - the hardware RAID controllers I've worked with store the RAID configurations both in NVRAM and on the disks, and this is what allows a controller to be easily swapped - the controller verifies it's stored configuration against the disk and if they don't match it will generate an error telling you what it found.  They are smart enough to know when the attached disks - so to speak - do not belong to it, and will point that out and give you the choice to load the configuration from the disks.

Regarding the "current backup export option" which I have never used - whilst I take your point on validating the data on re-import - does it not work when the user hasn't fiddled the data?  Is this not a case where the problem is the user?

fordem wrote:

Regarding the "current backup export option" which I have never used - whilst I take your point on validating the data on re-import - does it not work when the user hasn't fiddled the data?  Is this not a case where the problem is the user?

I work in software development and from a developer's point of view of course the problem is the user in this case. In fact the users is almost always the problem

But I don't think that's a reasonable answer for a consumer device in this situation. Users doing dumb things is expected and having things "go wrong" when users do dumb things is also often reasonable. But I don't think bricking a unit  to the point that it requires a warranty exchange is reasonable for a case where a user did nothing more than edit a text file produced by the webui and then re-uploaded a modified version of the file through the same UI. (No hardware changes, serial access, tampered firmware checksums, etc.)

End-user features like this should be robust enough to protect the device. All they would need to do it checksum the output config file so that basic tampering could be detected and rejected... (or have smarter validation that handled modified data but made sure it was actually a valid configuration; but just the checksum would be sufficient IMO).

-Jeff

Last edited by jdoering (2008-05-10 20:34:23)

jdoering wrote:

But I don't think that's a reasonable answer for a consumer device in this situation. Users doing dumb things is expected and having things "go wrong" when users do dumb things is also often reasonable. But I don't think bricking a unit  to the point that it requires a warranty exchange is reasonable for a case where a user did nothing more than edit a text file produced by the webui and then re-uploaded a modified version of the file through the same UI. (No hardware changes, serial access, tampered firmware checksums, etc.)
-Jeff

Let us not lose sight of the purpose of the configuration backup & restore utility - to backup a configuration so that it can be restored at a later date & time - why is the user editing it directly?

I think that what needs to be recognized here (in this forum) is that this device is a consumer product, intended for use by consumers who, for the most part, will not go digging around the innards - and that when we, in our quest for more, slip up and do something, that in hindsight looks like a really dumb thing (the proverbial awww-shiiit moment) that breaks the unit, that it didn't just break by itself, but rather that we broke it, and that warranty doesn't cover what we did.

jdoering wrote:

It's a consumer device; end-user mistakes should be correctable using hardware reset. Short of uploading hacked firmware; I believe that any bricking is the manufacturer's responsibility under warranty.
-Jeff

So - restoring a "hacked configuration" is in some way different to uploading hacked firmware?

I believe among the features in the later versions of fonz fun_plug is a script that allows the password changes to be written to NVRAM - is that in some way different to "uploading hacked firmware"?

No - what you are describing is not an end-user mistake - you're deliberately using the device in ways that the manufacturer never intended in an attempt to obtain increased functionality - which is the very essence of the term hack.

You hack it - using whatever means available - and in the process brick it and it should not be the manufacturer's responsibility

mig wrote:

jdoering wrote:

I noticed that disk ids are generated and stored on each disk and written to the DNS-323's flash-based minix filesystems.

What files are (or contain) the disk ids and were do you see these disk ids being stored?
Do you know what process is generating these disk ids?

I'm going to stop hijacking this thread; so I responded to the hd_magic_num question in a new thread: http://dns323.kood.org/forum/t2181-hd_m … tails.html

I'll also stop going on about config-restore here (while I certainly don't need to say any more on the topic; I will respond if relevant but only if a new thread is forked for the discussion)

I do think that the original question about transferring a RAID-1 set to a new DNS-323 is very important. Aside from standard disclaimers that you might loose all of your data; a confirmed working flow would be nice. I don't have all of my equipment setup to do such a test just this second - but I might take a look sometime soon.

(Edit) Sorry; I apparently misread fordem's earlier post and missed that he has confirmed RAID-1 "set" swapping. Good to know; I would like a solution that doesn't take additional drives (especially not a pair) but I don't know if that's possible on the DNS-323 without "hacking".

-Jeff

Last edited by jdoering (2008-05-11 03:35:23)

Thank you for your input.

I took the advice and bought me an external USB hard drive (Western Digital 750GB - $150 + tax, was something I could afford).

Now a whole new issue unrelated to my original question, but related to the backup advice….

     A quick rundown:     

OS = Windows Server 2003
DNS-323 = 2 x 750GB in RAID 1
USB HD = WD 750GB (attached to Windows Server 2003)

Reformatted the USB hard disk to NTFS with compression checked (why not, it doesn’t hurt).  I reformatted because from what I have read, but have never experienced, FAT32 has a file size limitation of 4GB, and I have DVD size ISO files.

Selected all folders on DNS323 and dragged it onto WD.  After a minute of preparing to copy, got a massage saying error can not copy file.  No mention of a file name.

So I started one folder at a time (from root).  First one went well.  Second one, got an error message saying it could not copy file (gave a name) from a certain folder, because it could not find the file!!  So went to the source folder, a nested folder not too deep, and sure enough, there was no file with that name.  What happened to that file?  If the file was not there to begin with, then where did the name come from and why was it copying a non-existing file?  I manually copied over the rest of the left over files and folders and...

This happened again, when I tried the next (third) folder.  Where do these files go?  Obviously they are there when attempting to copy, but when it gets to copy them, they seize to exist.

     Observations:

- I do not have a virus scanner, just ThreatFire that came with PC Tools Firewall.
- The two files that went missing (for lack of a better word), were .EXE files.
- File name and path does not exceed 255 characters.
- I am going to list the filenames and save them to a text file, before copying and see if I can find a pattern in the files that go missing.

It’s over 600GB and I really cannot go through it one folder at a time.

Any ideas?

Hoss,
   Do you have the DNS fun_pluged?  If so, there are a few things that may make backing this up easier.