Questions before I buy unRAID (warning - lots of text)

I am planning to make a media storage server, and have decided on either RAID6 or unRAID as the storage format.  Although I feel that I understand how RAID6 functions quite well, I have many questions about unRAID which I wanted to email and ask.  Some of these were briefly covered, or mentioned in passing on the forums, but I don't really feel as if they were clearly answered.

 

Firstly, the case I was considering (Norco server case from Newegg) holds 20 hot-swappable drives.  Is there a planned version of unRAID with support for up to 20 drives that will be released in the near future (next few months/year?)  If not, is it possible to run unRAID on top of another OS, such as Windows XP, and run a software RAID5 or Raid1 with the other few drives?

 

How difficult is the learning curve to use unRAID?  I am very familiar with Windows, but only know a few basic linux commands, and although I have dual-booted Ubuntu on my laptop in the past, I did not do much with the OS before that laptop suffered hardware failure.

 

Does unRAID Pro allow for 15 drives + a parity drive, or 16 drives + a parity drive (i.e. does the parity drive count against the maximum number of drives allowed)?

 

Reliability: One main problem with RAID is that, although the chance of two independent drive failures from an undegraded state is low, when the array enters a degraded state, there is a fairly high chance of a second failure while rebuilding the array (also referred to as 'stripe kill').  Even with RAID6, rebuilding an array is a tricky proposition.  With this in mind, how does the long-term reliability of unRAID compare to RAID6 (with any data loss being considered 'unreliable')?  I have seen basic statistics comparing it to RAID5, but in large arrays, RAID6 is much more practical.

 

Another reliability concern with RAID is starting and stopping the array - whenever a large RAID5 or 6 array is started up from powerdown, it's always a good idea to hold your breath until everything is running and spun up properly - does unRAID have similar issues with being shutdown and restarted?  (I assume not, but I'd better check in advance)  (I found one brief thread on this issue, but no clear answer)

 

One major plus of unRAID is its cheap price (no $1000 RAID card) and low startup cost (don't have to buy all drives at once) compared to RAID.  So much so, that if I go the unRAID route, I will probably make a second identical server to keep data backed up on.  Does unRAID provide any method by which I can nightly sync data between the two servers, or do I have to manually copy the data between each server?  What would be the easiest method of doing so if both are headless?  How can I transfer files between two headless unRAID servers?

 

unRAID uses ReiserFS.  Is there no GUI-based tool to recover deleted or partially corrupted data (such as r-studio for Windows)?  The only tools I found mentioned on the forums are commandline tools.  (on the plus side, you can recover data off one drive instead of needing to reconstruct an entire array...)  Additionally, is there any way to recover individual files that have been accidentally deleted?  The only recovery tools I could find were built around recovering an entire drive.

 

What happens if the flash drive itself is corrupted?  I know the  'unofficial' answer is that a new key will be sent to you.  However, does this mean that all data will be lost and the array will have to be reconstructed?  Or is making a backup of unRAID and the base filesystem as easy as copying all files except the .key file to a backup and copying them to a new key when needed?  Likewise, if I make a second identical server,

can I just copy the data from one key to another to setup unRAID and all installed packages and scripts or would I setup each computer separately?  Along these lines, can I backup the flash drive contents to an external USB hard drive or are the contents write-protected in some proprietary manner?

 

Is there way to set up unRAID as a dual parity system with two independent parity drives (similar to RAID6) so that the system can recover from two drive failures instead of one (or is this not generally an issue, since unRAID is not striped and stripe-kill is nonexistent)?

 

What is the false-positive failure rate and what is the best way to check whether a drive marked as failed is actually working?  Smartctl?  It seems like drives would be marked as failed often in the current system: power failures/spikes, loose cabling, and data overload, for example, could all cause failed writes? Has this happened to anyone?  (this question is touched on in various places, but nowhere is the false-positive failure rate specifically discussed)

 

As a practical matter, have those of you using unRAID found the drive failure rate greater for unRAID or RAID6?  Which puts more stress on the drives.  Admittedly, unRAID drives (except the parity) are usually spundown, and RAID drives are kept spun up, but does this actually make a difference?  (forum answers seem inconsistent)

 

Is there a way you can schedule parity checks or disk integrity checks for unRAID whenever it is not in use (rather than setting it in cron to run at a specific day and time?)

 

Can unRAID copy files from a CdROM drive on the server (I assume slackware comes with basic CdROM drivers?)  Can it burn files to a CD/DVDf not, what program is recommended for this? xcd-roast?  What about using external USB cd-rom drives and hard drives?

 

Does slackware (or unRAID) have a package manager and basic GUI that makes it easy to install and use programs onto the server?  Or is it all command-line based?

 

Is there any GUI tool to transfer files between server hard drives?  The website talks alot about midnight commander, but mc is rather un-ideal when it comes to reorganizing large numbers of media files with long filenames in different folders - dragging and dropping would be much more efficient.  This is very confusing: the forums talk about how you can use a mouse through putty, but then only give commandline instructions for transferring files – what good is a mouse without a GUI?  Along these lines, can you use a VNC client with the server (assuming there's some GUI to use it with)?

 

What is the easiest way to transfer files to and from the server?  User shares under Windows Explorer?  FTP access (this would probably be the easiest)?  MOST importantly, is there some sort of automatic hash or .sfv calculation and checking in such transfers (especially FTP transfers?) to prevent corruption in large file transfers (I occasionally have file corruption over FTP when transferring large video files - not an issue for temporary viewing but a real probelm for archiving)?  What file transfer method would you suggest using and why?

 

Will unRAID support the use of hot spares anytime in the future (like most RAID cards do)?  This is especially important in light of the fact that it is easy to overlook a failure in unRAID (unless you check the status screen daily)

 

Can you mount NTFS drives natively in unRAID(or the Slackware it runs on) to transfer files to the ReiserFS drives used by unRAID, or do you need extra programs/drivers?  (I read that you can install a driver in Windows to mount unRAID volumes as read-only in Windows, but haven’t read much about the other way round)

 

unRAID is obviously slower than striped RAID, which is okay since I'd mostly be using it for storage.  However, is it fast enough to stream 1080p HD video over a gigabit ethernet connection?  As far as I can tell, most forum members seem to be getting 20-30 Mb/s read/write off unRAID, although I can’t find much data on network transfer speeds and how fast they need to be to stream HD video.

 

Can unRAID interface properly with case hard drive lights as various RAID setups can such that any bad hard drive would give a red light next to the corresponding hard drive on the case front?  (Or is the web GUI the only way to check for a bad drive?)

 

Does the array ever need to be defragmented? (it seems more like traditional storage than the striping used in raid, which does not need defragmenting)  I read that the ReiserFS filesystem is alot less prone to fragmentation than NTFS, but surely after several years of storage and moving files, fragmentation would be an issue?  If so, does unRAID include a defragmentation utility?  Can you use reiserfsck?  If 

not, then what?

 

What exactly does reiserfsck do?  Smartctl seems to give error rates and temperatures for the drives only, correct?  Is reiserfsck the equivalent of windows checkdisk?  What does it do that smartctl doesn't?

 

When the array configuration data is reset, what exactly happens?  Is all the data in the array lost and all drives reformatted, or is it just that the parity drive data is declared invalid and a new parity constructed based on the remaining working drives?  Different forum topics seem to imply different things.

 

If the array has extra unused storage space greater than the space available on a failed array drive, is it possible to simply pull the failed drive and reconstruct the array to a nondegraded state with the fewer number of drives without losing any data?  Or will the data on the failed drive have to be reconstructed onto a new unused drive, then copied over before the array can be used again?

 

How long will adding a new disk take with a 1.5 Tb disk (the largest current disk)?  Are we talking minutes, hours or days?  (Parity checks seem to take 5-10 hours, but the time to rezero a disk has been given as anywhere from an hourish for a TB disk to over a twenty hours for a 200 GB disk – something’s a bit inconsistent there…)

 

Does the Slackware distribution bundled with unRAID come with a GUI (similar to Ubuntu or Kubuntu, say)?  Also, how difficult is it to run the server headless?  Can I access the GUI with a VLC program?  If no GUI, I assume I just connect to a command prompt via putty?  Will I need a monitor to initially install unRAID, or can I install the OS and configure it all from a putty connection?  (Some of these questions are repeats, but they seem to fit well here too)

 

Can I run a media manager on my media center PC that I can use to access and sort files on the share based upon independently entered characteristics (i.e. by type, extension, encoding, quality, resolution, genre, etc.)?  For example, is unRAID compatible with MainLobby, Media Portal, or other programs running on a remote PC?  I assume I just point the media center to the remotely hosted files?  Can I set a MAC address restriction on the media center PC via the unRAID configuration interface that would let the media center access these files without a password (but make other computers still use a user access password)?

 

What exactly am I paying for?  Is unRAID a modification of the linux kernel, or a closed-source proogram than can be run on Slackware, or what?  What exactly am I buying and why (considering that linux is open-source, with a very open license)?  I don’t think I’ve ever been charged for a program that ran on a *nix distribution before, so was just wondering…

 

What is the fewest number of drives that can be used to build an unRAID array?  2?  (1 drive and 1 parity) (note: I understand that a two drive system is rather pointless, but it would be a good place to start expanding from as I fill up drives)  Will unRAID refuse to work with fewer than 3?

 

Is there any auto-update utility or any way to update the unRAID version or Slackware version without having to rebuild the array, redownload packages, rewrite scripts, etc. (Do I just copy the new files over the old ones - are the config files stored separately?)  How does upgrading Slackware work?  (Again the forums touched on this, but were confusing).

 

The forums mentioned that if bad sectors existed on a disk, and a reconstruction was being accomplished due to a failed (other) disk, data could be lost.  Thus, regular parity checks are recommended.  Is parity checking the only way to check for and reassign bad sectors?  Will Smartctl do this as well?  Reiserfsck?  (This question ties back into the checkdsk equivalent question earlier) - I'm a bit confused as to the differences in preventative maintainance performed by smartctl, parity check, and reiserfsk (there seems to be a bit of overlap, in the area of bad sectors in specific), and which you need to run and how often?

 

Tying into the previous question, how do you schedule tests?  Is there a GUI that comes with unRAID or is the current option (still) to write cron scripts for each test you want to schedule?  Does unRAID come with cron and documentation that explains how to use cron, or should I just copy/paste scripts provided on the forums?

 

A 'cache drive' was mentioned in the forums once, but no details given - what do you mean by using a cache drive to increase performance?  Does this count against the 16 disk limit?  What exactly is it for and how do you set one up if necessary?

 

I have an APC, and would want to have the computer autoshutdown if it had to run on APC power for more than a minute or two.  There is a many-pages long, confusing forum topic on this issue, which seems to something being currently worked on in various Slackware/unRAID versions.  Obviously, the software that came with the APC will not run on slackware.  I read about a possible solution on the forum, but it seems very involved.  Is it still necessary to recompile the slackware kernel (which I have no idea how to do) and install additional drivers to enable APC powerdown support?  How hard is this to set up?

 

I was originally going to email this to the developers, but figured I'd get as much answered as I could on the forums first, then email any additional questions before making a decision.

 

Thanks to all respondents in advance for your help, and I apologize if these questions have been already answered (I've spent many hours browsing the forums - yes, I HAVE read all the FAQS/documentation as well - and some answers I could not find or were confusing).  Most of my questions can likely easily be answered in a few words.  Some of them (such as the APC question) have been answered, but not concisely, or inconsistent answers were given.  In these cases, please do no simply post a forum link, as I have likely already read the topic and am still confused.

 

I have a fairly good knowledge of computer hardware and some programming skills (read: it’s been awhile since I had two write any code).  I have an extensive knowledge of Windows, but little practical linux experience (read: I know what I want to do, and often if it can be done, but not necessarily the programs or commands to actually do it in *nix).  I know how to configure network security, and although I’m pretty sure I can learn how to write my own scripts if I need to, if I would have to spend days writing and debugging scripts to get unRAID to do what I want it to, I might as well go with RAID6, especially as each new script I write, debug, and primarily use myself would be another potential failure point in what I want as a relatively reliable data archive.  I am willing to spend a week or two of spare time to set this array up, but want it to work without much intervention on my part afterward, aside from adding the occasional drive overnight (hence, many of the above questions concern the stability, reliability, accessibility, and automation of unRAID, rather than ‘how it works’).

 

Sorry for the insanely long post, but I think this sums up pretty much all the concerns I have about unRAID.  If someone familiar with the system could take 15 minutes and type up a couple words in answer to each question point by point, it would be much appreciated.

 

Thanks again,

Zithras

 

After your very thorough analysis of the information available on both this forum and the WIKI, (I won't post any links as you requested), about the only question I didn't see was whether or not a FREE version were available so that you could test whether unRAID would meet your requirements.  The good news is that there is!!! (again, no links.)

 

Welcome to the unRAID community.   

 

Please don't interpret my failure to answer any of your specific questions with specific answers to be 'noob bashing'.  This is one of the friendliest communities that you will ever find.  But, frankly, after fully reading your list of questions the best advice I can offer is for you to simply install unRAID on some spare hardware and play with it.  There just is no better way to answer the majority of your questions.  I believe you might even discover something in the process that you'll be able to share with the rest of us.

 

I can tell you that before I jumped in I was very unsure as well.  Now, I'm addicted and my unRAID server is one of my most valuable pieces of hardware (I won't embarrass myself by mentioning just how many machines I have running in my home).

 

I will say that some of your requirements will require going a bit beyond unRAID functionality that is GUI configurable (i.e. rsync might be a logical solution to some of your requirements).  To this end, I'd suggest installing unMENU once you get unRAID up and running.  It, among other features, offers a great way to manage package installation.  After that, I recommend adding the myMAIN plug-in to unMENU (also very cool).

 

I could go on, but I'd just be getting ahead of myself. 

Sorry for the insanely long post, but I think this sums up pretty much all the concerns I have about unRAID.  If someone familiar with the system could take 15 minutes and type up a couple words in answer to each question point by point, it would be much appreciated.

 

Thanks again,

Zithras

I plan on answering this post, question by question, but I'll do that *after* I have my morning coffee.   It will take me much more than 15 minutes to answer, as I hope to put links to threads and specific items in the wiki that answer your questions....  I expect this will help a lot of others, you just happened to put it all in one fairly complete list of questions.   

 

As already said, your needs might be satisfied by unRAID, but there are times when a command line entry of a linux/unix command must be made.  Those are not the normal operations, but when needed there is often no GUI equivalent.   

 

You recognize that unRAID is built on Linux, specifically "Slackware"  It is a stripped down version of Slackware 12, leaving off the X-windows based "desktop" and most extras. It does fit on a small flash drive. I think it was originally shipped on a 128Meg flash drive... although you would find that size almost impossible to find today.

 

I hear the coffee pot behind me is finished brewing...   

 

Joe L.

 

Reliability: One main problem with RAID is that, although the chance of two independent drive failures from an undegraded state is low, when the array enters a degraded state, there is a fairly high chance of a second failure while rebuilding the array (also referred to as 'stripe kill').

 

As you know unRAID is not striped.  If you lose 2 drives, you only lose data from those 2 drives.  The data on all other drives is fine.  Plus, data recovery from a failed unRAID drive is a simple matter for any data recovery service, while recovery from a striped RAID array is practically impossible.

 

Another reliability concern with RAID is starting and stopping the array

 

Because it is not striped, there is much less trepidation with startup and shutdown.

 

Does unRAID provide any method by which I can nightly sync data between the two servers,

 

No.  The standard Linux tool "rsync" is used for that by most people.

 

Geez... you have 20 TB of data, *and* want to mirror it to an identical server?  What situation would you have such a large amount of such critical data?

 

Is there no GUI-based tool to recover deleted or partially corrupted data

 

No.  Read my lips... GUI is bad... command line is good.  Break your addiction.

 

Additionally, is there any way to recover individual files that have been accidentally deleted?  The only recovery tools I could find were built around recovering an entire drive.

 

No... you recover the whole drive.  ANd since you are planning to mirror the whole server, you would do better to just copy it from the mirror than rebuilding a drive.

 

What happens if the flash drive itself is corrupted?

 

The entire contents of the flash is copyable and can be backed up.

 

Is there way to set up unRAID as a dual parity system with two independent parity drives

 

Development of a dual parity implementation for unRAID has been announced by Tom.

 

As a practical matter, have those of you using unRAID found the drive failure rate greater for unRAID or RAID6?

 

Comparing unRAID to any type of striped RAID is comparing apples and oranges.  First, the dataset of people who have extensive experience with both is so small as to provide nothing but anecdotal data.  Second they are different markets, for different purposes.  Third, RAID failures are much more serious than unRAID failures since in a striped set, you are one step from total disaster and 100% data loss with practically zero chance of data recovery from a data recovery service.  Recovery of a data from a failed unRAID drive is actually better (more likely to recover data intact) than recovery of a desktop drive since unRAID is not going to screw around with the drive and continue to use it and potentially cause more damage.

 

Don't get me wrong, no one likes coughing up a few hundred buck for a drive recovery.  But anyone in this business long enough will need it once or twice.  It works.  It's your insurance policy, but you don't have to pay for it until you need it.  But with a striped RAID, unless you are using one of the handful of systems that the data recovery service can handle, you are SOL.

 

Is there a way you can schedule parity checks or disk integrity checks for unRAID whenever it is not in use (rather than setting it in cron to run at a specific day and time?)

 

Nope.

 

Can unRAID copy files from a CdROM drive on the server

 

Yup... you just have to mount the CD.

 

Can it burn files to a CD/DVDf

 

Not unless you install some software to do so.  But why would you?  Burn Cds from unRAID data on your desktop CD drive.... just map the unRAID drive.

 

Does slackware (or unRAID) have a package manager and basic GUI that makes it easy to install and use programs onto the server?  Or is it all command-line based?

 

Again, embrace the command line.  Installing packages takes 2 commands, and you just cut and paste the file name/URL you want to install.

 

what good is a mouse without a GUI?

 

Copy and paste.  And in many text applications, such as MC, the mouse works just like in a GUI... click on buttons, etc., it is just that the buttons are made from text, not pictures.

 

Is there any GUI tool to transfer files between server hard drives?

 

Yes, it is called "windows" from a company called Microsoft.  Or Mac OS X or even Linux.  When you have mapped the drives to your desktop, you can use whatever GUI you have on your desktop to move files around.  Samba is quite smart about this, and will simply rename or move files rather than copying, and even when it does copy, it copies locally and not over the wire when the copy is within the same server.

 

Along these lines, can you use a VNC client with the server (assuming there's some GUI to use it with)?

 

I use VNC to remote control KDE on my unRAID server... but to do so, you have to install a complete Slackware distro, and rebuild your own kernel.  See this wiki article:

 

http://lime-technology.com/wiki/index.php?title=Methods_of_user_customization

 

Can you mount NTFS drives natively in unRAID

 

Yup.

 

However, is it fast enough to stream 1080p HD video over a gigabit ethernet connection?

 

Short answer:  Yup.

Better answer:  What is the bitrate of your 1080P stream?  A raw VC1 with an insane bitrate (over 250 Mbps) could cause problems.  But if you store HD in h.264 or MP4, you'll have no problems.

 

 

Can unRAID interface properly with case hard drive lights

 

Not intelligently.  If you have a drive cage with lights for on/off busy/idle, etc., they will work.

 

Does the array ever need to be defragmented?

 

No.

 

What exactly does reiserfsck do? 

 

Read the man page for reiserfsck.  Google "man reiserfsck"

 

Smartctl seems to give error rates and temperatures for the drives only, correct? 

 

Read the man page for smartctl.

 

Is reiserfsck the equivalent of windows checkdisk?

 

 

It is similar in concept.

 

What does it do that smartctl doesn't?

 

Compare the two man pages

 

When the array configuration data is reset, what exactly happens?  Is all the data in the array lost and all drives reformatted, or is it just that the parity drive data is declared invalid and a new parity constructed based on the remaining working drives?

 

Neither.  Drives that are valid unRAID disks will still have their data, as long as you have not reassigned it to be parity.

 

If the array has extra unused storage space greater than the space available on a failed array drive, is it possible to simply pull the failed drive and reconstruct the array to a nondegraded state with the fewer number of drives without losing any data?

 

When the drive fails, you can still access its data since unRAID will transparently reconstruct it on the fly as you attempt to read it, so if you have the free space, you can copy the data to another location, and when done, stop the array, delete the failed drive from the configuration, and restart it.  That will kick off a parity rebuild.

 

How long will adding a new disk take with a 1.5 Tb disk

 

Depends on the system.  Clearing (writing zeros to the whole disk) takes the time.  You can clear a new disk outside the array with the script from Joe L.

 

Does the Slackware distribution bundled with unRAID come with a GUI

 

No.

 

how difficult is it to run the server headless?

 

They only time being headless is a problem, is if you need to futz with the mobo BIOS, or net connectivity is out.  Otherwise telnet works just fine.

 

Will I need a monitor to initially install unRAID, or can I install the OS and configure it all from a putty connection?

 

Some people have to futz with the mobo BIOS to get the mobo to boot properly from the unRAID USB stick.  After that, you'r home free.

 

For example, is unRAID compatible with MainLobby, Media Portal, or other programs running on a remote PC

 

If they support Samba, yes.

 

What exactly am I paying for?  Is unRAID a modification of the linux kernel, or a closed-source program than can be run on Slackware, or what?

 

unRAID is 1) a custom, minimalist Slackware build, with 2) a GPL's modification to the standard MD driver and 3) a proprietary management interface.

 

 

What is the fewest number of drives that can be used to build an unRAID array?

 

One.  Single data drive w/o parity.

 

Is there any auto-update utility or any way to update the unRAID version or Slackware version without having to rebuild the array, redownload packages, rewrite scripts, etc.

 

There is an interactive script you can run, from some guy named "Bubba"

 

Is parity checking the only way to check for and reassign bad sectors?

 

No, but it is the simplest to do the whole system in one fell swoop.

 

Does unRAID come with cron and documentation that explains how to use cron, or should I just copy/paste scripts provided on the forums?

 

unRAID comes with cron but unRAID does not include man pages.  Google "man <command name>" for them.

 

I have an APC, and would want to have the computer autoshutdown if it had to run on APC power for more than a minute or two.

 

The posts on that topic are deceptively complex... it is actually quite simple to install the apcupsd package and configure it.

 

The forums mentioned that if bad sectors existed on a disk, and a reconstruction was being accomplished due to a failed (other) disk, data could be lost.

 

Same thing happens with striped RAID... but the results for striped raid are *disaster*...all data unreachable.  But hey, you said you are going to mirror the whole server so no problem... right?

 

 

 

 

I plan on answering this post, question by question, but I'll do that *after* I have my morning coffee.  It will take me much more than 15 minutes to answer, as I hope to put links to threads and specific items in the wiki that answer your questions....  I expect this will help a lot of others, you just happened to put it all in one fairly complete list of questions. 

 

Would it be worthwhile to update the FAQ with some of these questions.

Seems the time involved in writing the text and posting links may be useful in enhancing the FAQ and posting that link.

 

That's allot of questions. It will take some time just for me to read them LOL!!!

 

Thanks Joe CooL

I am planning to make a media storage server, and have decided on either RAID6 or unRAID as the storage format.  Although I feel that I understand how RAID6 functions quite well, I have many questions about unRAID which I wanted to email and ask.  Some of these were briefly covered, or mentioned in passing on the forums, but I don't really feel as if they were clearly answered.

Here goes... hopefully, some of your questions will be answered in a bit more detail

Firstly, the case I was considering (Norco server case from Newegg) holds 20 hot-swappable drives.  Is there a planned version of unRAID with support for up to 20 drives that will be released in the near future (next few months/year?)

Yes, there is a planned version that will support more drives.  See here  Note that even though the case may be hot-swapable, in general Linux is not, and unRAID is not "hot-swapable." You will need to power cycle, or (at a minimum) reboot to add a physical drive to the array.

If not, is it possible to run unRAID on top of another OS, such as Windows XP, and run a software RAID5 or Raid1 with the other few drives?

Some people have run unRAID under vmware on another OS, some have run vmware on unRAID to run windows.  Both are very not for the beginner, as they do require you to get involved with adding packages of software and merging unRAID with other slackware and vmware software.  See here and Here

How difficult is the learning curve to use unRAID?  I am very familiar with Windows, but only know a few basic linux commands, and although I have dual-booted Ubuntu on my laptop in the past, I did not do much with the OS before that laptop suffered hardware failure.

It is very easy to use unRAID.  You will need to know very little about Linux.  If you were able to configure a dual-boot laptop, you should be fine.  Lots of help here in the forum when something at the Linux command line level is needed.   Lots of "scripts" and utilities exist to help you know how the server is performing.

Does unRAID Pro allow for 15 drives + a parity drive, or 16 drives + a parity drive (i.e. does the parity drive count against the maximum number of drives allowed)?

Currently, the limit is 15 data drives plus a single parity drive, plus a "cache" drive in the array.  You can also currently have as many drives as you like outside of the array.  These additional drives are not managed through the Lime-technology supplied management interface, but through either the unMENU add-on Alternate Improved Interface, or via Linux command-line commands.  These additional drives are not protected by parity, nor can they be made into a raid array of their own, but they are not counted as part of the 17 drives unRAID does manage (15 data, 1 parity, 1 cache).

Reliability: One main problem with RAID is that, although the chance of two independent drive failures from an undegraded state is low, when the array enters a degraded state, there is a fairly high chance of a second failure while rebuilding the array (also referred to as 'stripe kill').  Even with RAID6, rebuilding an array is a tricky proposition.  With this in mind, how does the long-term reliability of unRAID compare to RAID6 (with any data loss being considered 'unreliable')?  I have seen basic statistics comparing it to RAID5, but in large arrays, RAID6 is much more practical.

The current version of unRAID suffers from the same danger of a second drive failing concurrently while a first drive has failed.  UnRAID is better than RAID-5 in that if this were to occur, you could still get to all the data on any of the remaining drives that were still working.  In RAID-5, you would lose all the data on all the drives if two were to concurrently fail.    But... at some time, hopefully, Tom will keep to his plans to release the version of unRAID as described in this post that will support 2 parity drives and the ability to recover from 2 drives failing concurrently.

Another reliability concern with RAID is starting and stopping the array - whenever a large RAID5 or 6 array is started up from powerdown, it's always a good idea to hold your breath until everything is running and spun up properly - does unRAID have similar issues with being shutdown and restarted?  (I assume not, but I'd better check in advance)  (I found one brief thread on this issue, but no clear answer)

If your power supply is not up to the task of spinning up a large number of drives at one time, it will have issues.  You are advised to get a supply that has a single 12 volt rail rather than multiple 12 volt supply rails.  See Here and Here

One major plus of unRAID is its cheap price (no $1000 RAID card) and low startup cost (don't have to buy all drives at once) compared to RAID.  So much so, that if I go the unRAID route, I will probably make a second identical server to keep data backed up on.  Does unRAID provide any method by which I can nightly sync data between the two servers, or do I have to manually copy the data between each server?  What would be the easiest method of doing so if both are headless?  How can I transfer files between two headless unRAID servers?

unRAID includes the "rsync" utility.  It allows you to keep two machines in sync.  (I've never used it, as I have no similar need to keep two Linux disks synchronized)  It is a command line based utility, but it can be scheduled using the chronological scheduler built into Linux.  Again, there is no GUI interface, but plenty examples of how to use it on the web.  It is not specific to unRAID, or even to Linux, as there are Windows "rsync" programs available too.

unRAID uses ReiserFS.  Is there no GUI-based tool to recover deleted or partially corrupted data (such as r-studio for Windows)?  The only tools I found mentioned on the forums are commandline tools.  (on the plus side, you can recover data off one drive instead of needing to reconstruct an entire array...)

Actually, you don't recover data, you simply access the stand-alone normally. (worst case, put it in a different PC, boot up the unRAID flash drive, assign the one data drive, do not assign any parity drive, and you have access to that one drive on your LAN.)

  Additionally, is there any way to recover individual files that have been accidentally deleted? 

Short answer, Linux has no "un-delete".  Longer answer, If your files are not in the top directory in the file-system, and if you have not re-used the freed data blocks on the disk with subsequent writes, you can use the reiserfsck command line utility to recover files that were just deleted by asking it to rebuild the file-system-tree. See Here.

The only recovery tools I could find were built around recovering an entire drive.

True, you cannot "recover" a single file.  But, it should do what you want.

What happens if the flash drive itself is corrupted?  I know the  'unofficial' answer is that a new key will be sent to you. 

If the flash drive is still writable, you can just re-load it.  If not, a new flash drive can be obtained as you described.  Regardless, you can boot an un-registered copy of unRAID and get access to any two of your data drives. (Just don't assign a parity drive)  Once you get the new registration key, just assign the remaining drives.  The data will still be there, just as you left it.  You do not need to reconstruct your array.  You will probably be forced to re-calculate parity, but the unRAID management utility will do this for you when you assign the drives to the array.

However, does this mean that all data will be lost and the array will have to be reconstructed?

Just need to re-calculate parity, and this will be done for you when you assign the drive to the config stored on the new flash drive.

  Or is making a backup of unRAID and the base filesystem as easy as copying all files except the .key file to a backup and copying them to a new key when needed? 

That is EXACTLY what it involves.

Likewise, if I make a second identical server,

can I just copy the data from one key to another to setup unRAID and all installed packages and scripts or would I setup each computer separately?

You can copy MOST files, but unless the second server has exactly the same disk drives with exactly the same serial numbers (unlikely) you will not want to copy the disk.cfg, or super.dat files from one server to the other.  Nor would you want to copy the network.cfg file, as it has the server name, and possibly the IP address if you use a static IP address.

  Along these lines, can I backup the flash drive contents to an external USB hard drive or are the contents write-protected in some proprietary manner?

The contents are not protected in any way.  You can copy the contents anywhere you like.

Is there way to set up unRAID as a dual parity system with two independent parity drives (similar to RAID6) so that the system can recover from two drive failures instead of one (or is this not generally an issue, since unRAID is not striped and stripe-kill is nonexistent)?

Answered earlier.  See here

What is the false-positive failure rate and what is the best way to check whether a drive marked as failed is actually working?  Smartctl?  It seems like drives would be marked as failed often in the current system: power failures/spikes, loose cabling, and data overload, for example, could all cause failed writes? Has this happened to anyone?  (this question is touched on in various places, but nowhere is the false-positive failure rate specifically discussed)

If a drive is marked as "failed" it is because repeated attempts to write to it by the OS have failed.  (Before the unRAID driver is notified of a "write" failure, the lower level device driver has tried and re-tried)  In my experience reading this forum, if a drive is marked as failed, it is not "false-positive"  It might be intermittent, but not false.  As you said, a loose cable can cause this, but I've not heard of a power failure do it, nor of any "data overload"  (A drive filled with data does not fail in the same way.  It will not be marked as failed by unRAID)  A poor quality cable (Older 40 conductor, or excessively long, and/or "round" IDE cables) can also cause intermittent errors, but usually they will occur on reads of data.  For the most part, unRAID arrays today are SATA based, so these errors are far less common.  More common is a cable that gets un-plugged.

As a practical matter, have those of you using unRAID found the drive failure rate greater for unRAID or RAID6?  Which puts more stress on the drives.  Admittedly, unRAID drives (except the parity) are usually spundown, and RAID drives are kept spun up, but does this actually make a difference?  (forum answers seem inconsistent)

Can't help you there.  I'd expect the drive failure rate to be about the same, depending on usage profile.  A drive will wear out its bearings over time... All will. So a spinning drive incurs wear, a non-spinning drive does not.  The process of loading and un-loading the drive heads involves some wear too.  I personally don't know if one is worse than the other.  All drives fail...  In fact, there are only two kinds of hard-drives... those that have already crashed, and those that have not yet crashed, but will, just give it more time  .

 

Is there a way you can schedule parity checks or disk integrity checks for unRAID whenever it is not in use (rather than setting it in cron to run at a specific day and time?)

No, but parity checks do not normally affect the use of the server in any way.  You can still read and write to the array while they are occurring.  Many of use have a monthly "parity" check scheduled.  This is NOT actually to check parity, but instead to give the hard-disk's SMART firmware the chance to identify any disk blocks that are un-readable.

 

Can unRAID copy files from a CdROM drive on the server (I assume slackware comes with basic CdROM drivers?)  Can it burn files to a CD/DVDf not, what program is recommended for this? xcd-roast?

Although Slackware certainly comes with a CD driver, I do not think unRAID has one, remember, it is a stripped down version of Slackware.  As such, it has no ability to burn CDs.  Remember, it is basically designed for network storage.

  What about using external USB cd-rom drives and hard drives?

External USB hard drives are easy.  I use them frequently.  You can either use command line commands to mount them, or, install the unmenu interface.  It allows you to mount and share the external USB drives with ease.  

Does slackware (or unRAID) have a package manager and basic GUI that makes it easy to install and use programs onto the server?  Or is it all command-line based?

The Lime-technology supplied software has no package manager.  The unMENU improved interface has a package manager.  It will allow you to download and install those packages where a configuration file has been created.  See here

Is there any GUI tool to transfer files between server hard drives?  The website talks alot about midnight commander, but mc is rather un-ideal when it comes to reorganizing large numbers of media files with long filenames in different folders - dragging and dropping would be much more efficient.  This is very confusing: the forums talk about how you can use a mouse through putty, but then only give commandline instructions for transferring files – what good is a mouse without a GUI?

mc uses a character based GUI...  But since you want to drag and drop, you can do that from your windows PC using file explorer, it is just not as efficient as copies and moves done locally on the server at the command line level.   Look for a future unRAID plug-in module to help with some of those moves.

  Along these lines, can you use a VNC client with the server (assuming there's some GUI to use it with)?

There is no VNC server supplied since the interface is via telnet and there is no need for VNC to see a graphical desktop. I am not aware of anybody who has installed one.  You  can of course, install unRAID over a full slackeware distribution and have it all, but this is a very advanced task, and you are probably not a good candidate with your level of "appliance operator" skills to do it.  See here

What is the easiest way to transfer files to and from the server?  User shares under Windows Explorer?

This is by far the easiest.  Be aware, "User Shares" are different than "Disk Shares"  Both present themselves (when enabled) as shared drives on the LAN. The "disk" shares as "disk1, disk2, disk3, etc... one each per physical disk.  The "User Shares" are one each per top level directory on the physical disks.  I have a "Movies" directory on each of my physical disks.  The consolidated set of "movie files" in those "Movies" directories is presented to the LAN as a single folder with all my movies available for my media players.

  FTP access (this would probably be the easiest)?

Not really, as many people have never used FTP.

  MOST importantly, is there some sort of automatic hash or .sfv calculation and checking in such transfers (especially FTP transfers?) to prevent corruption in large file transfers (I occasionally have file corruption over FTP when transferring large video files - not an issue for temporary viewing but a real probelm for archiving)?

I am not aware of any hash checking or function.  Best I can offer is for you  to do an MD5 checksum on both the windows and Linux file.  This will involve the command line.  However... TCP/IP checks the contents for you.  Unless your hardware is flaky (bad memory, etc) you should not have any file-corruption when transferring a file using SAMBA over the LAN.

   What file transfer method would you suggest using and why?

I just use the shared drives on the LAN... easiest, and plenty fast enough for my use.   Once a parity drive is assigned to the array write speed is not limited by the LAN, but by the need to do 4 I/O operations for each block written to the server.

Will unRAID support the use of hot spares anytime in the future (like most RAID cards do)?

Unlikely... It has not been mentioned on the "Laundry list" of things Tom intends to impliment.

This is especially important in light of the fact that it is easy to overlook a failure in unRAID (unless you check the status screen daily)

Many of us have added a script to provide an e-mail notification when the array status is abnormal.  See Here.  There is no need to look at the management interface every day.

 

I'll continue the answers in the next post... Apparently, a single message in this forum cannot be over 20000 characters... and you asked a lot of good questions (or at least put them all in once place)

 

Joe L.

Continuing with answers...

Can you mount NTFS drives natively in unRAID(or the Slackware it runs on) to transfer files to the ReiserFS drives used by unRAID, or do you need extra programs/drivers?

Yes, easily...  There is a built-in read-only driver, you can read how to use it here, or you can install a read/write driver and have full access to the NTFS drive.  See Here . or easiest, use the "Disk Manager" plug-in I created in the unmenu Improved interface.  With it you can mount, and share the NTFS drive(s).  Then, a copy can be done using any tool or command you like.

(I read that you can install a driver in Windows to mount unRAID volumes as read-only in Windows, but haven’t read much about the other way round)

Well, now you can follow the links and read about the "other way around"

unRAID is obviously slower than striped RAID,

For "writing" yes, but for "reading," not really, at least not when accessed over the LAN

which is okay since I'd mostly be using it for storage.  However, is it fast enough to stream 1080p HD video over a gigabit ethernet connection?  As far as I can tell, most forum members seem to be getting 20-30 Mb/s read/write off unRAID, although I can’t find much data on network transfer speeds and how fast they need to be to stream HD video.

Many people do very well with full bitrate 1080p HD material.  Some even do it over 100MB Lan, as that is as fast as their media player supports.

Can unRAID interface properly with case hard drive lights as various RAID setups can such that any bad hard drive would give a red light next to the corresponding hard drive on the case front?  (Or is the web GUI the only way to check for a bad drive?)

The web-GUI is the only "stock" way.  However, you can install scripts for e-mail notification, or LAN based notification via a "Yac" client

Does the array ever need to be defragmented? (it seems more like traditional storage than the striping used in raid, which does not need defragmenting)  I read that the ReiserFS filesystem is alot less prone to fragmentation than NTFS, but surely after several years of storage and moving files, fragmentation would be an issue?  If so, does unRAID include a defragmentation utility?

As far as I know, reiserfs does not have a "defragmentation" utility.  As you might expect, it one was really needed it would have been created by the Linux developer community.  As a "media server" this is not really a huge iissue in any case, as most files will be only written once, and never updated.

  Can you use reiserfsck?  If 

not, then what?

What exactly does reiserfsck do?

No, reiserfsck is roughly equivalent to chkdisk in windows.  It checks, reports on, and fixed file-system structure damage, but does nothing to change the files themselves.

  Smartctl seems to give error rates and temperatures for the drives only, correct?

Yes, smartctl reports on the hard drive's S.M.A.R.T FIRMWARE error statistics.  It does not care what is on the disk physically, nor care if there is any file-system at all.  It just knows about reads and writes, and how well the disk hardware is performing.

Is reiserfsck the equivalent of windows checkdisk?

Yes

What does it do that smartctl doesn't?

It checks the reiserfs file-system (and can also repair it if damage is found)  It cares about the data in a specific partition file-system, not about how high the drive's head is over the disk surface, or how long it took to spin up, or drive temperature, etc... (SMART attributes)

When the array configuration data is reset, what exactly happens?

Good question... When you press the button labeled "restore" and it resets the array configuration all that happens is that the "/boot/config/super.dat" file is re-named to be "/boot/config/super.old"  When that occurs, the unRAID array is forced to build a new superblock (the data in the super.dat file is known as the superblock).  That file contains the drive sizes, the model and serial numbers if the drives, and the array "state" (STARTED, STOPPED, etc) as well as the date of the last parity calc and the parity status.  The new "super.dat" file will be constructed using the information in the disk.cfg file in combination with the drives listed in it that are working.  (In other words, a defective drive, even if in the disk.cfg file, would NOT be part of the newly initialized array)

    Is all the data in the array lost and all drives reformatted?

No, no "data" is lost, nor any drive "reformatted"

or is it just that the parity drive data is declared invalid and a new parity constructed based on the remaining working drives?  Different forum topics seem to imply different things.

It is just the parity drive considered invalid, so it will be re-written with a completely new set of data based on the working and currently assigned drives in the disk.cfg file

If the array has extra unused storage space greater than the space available on a failed array drive, is it possible to simply pull the failed drive and reconstruct the array to a nondegraded state with the fewer number of drives without losing any data?  Or will the data on the failed drive have to be reconstructed onto a new unused drive, then copied over before the array can be used again?

Good question.  I did something similar to this recently...  You can do the following when one drive has failed:

 

1. Copy all the data you wish from the "failed" drive to others in the array with spare space.  The "failed" drive's contents will be re-created by reading the parity drive in combination with all the other drives.

2. Stop the array, go to the "devices" page and un-assign the failed drive.

3. Go back to the main management page, check the checkbox under the button labeled "restore" and press it.  It will then immediately start the process of calculating parity on all the remaining drives, leaving out the failed drive, since it is no longer assigned to the array.  This will take some time, during which you have no parity protection, but then with the failed drive you had no protection from a second drive's failure anyway, so to me it was a no-brainer, even though it took about 12 hours on my array to rebuld parity. (I have an older IDE based server with a lot of disks)  Once parity was calculated, I had full protection against a second drive's concurrent failure... since the first failed drive was not in the array any longer.

How long will adding a new disk take with a 1.5 Tb disk (the largest current disk)?  Are we talking minutes, hours or days?  (Parity checks seem to take 5-10 hours, but the time to rezero a disk has been given as anywhere from an hourish for a TB disk to over a twenty hours for a 200 GB disk – something’s a bit inconsistent there…)

I've written a tool to pre-clear a disk and burn-it-in (exercise it so early failures can be identified)  It takes about 4 hours to clear a 1TB drive on my server.  It takes about 10 hours for a complete pre-read/zero/post-read cycle.  It would probably take about a third more time for a 1.5TB drive.

 

If you do not pre-clear a drive, your array will be off-line while the drive is cleared during the process of adding it to your array..  If you do pre-clear a drive, it can be assigned and formatted when added to the server in minutes.  Anybody with a 200 hour clearing process has bad hardware... (usually the drive has resorted to PIO access as a result of read failures and slowed way down)

Does the Slackware distribution bundled with unRAID come with a GUI (similar to Ubuntu or Kubuntu, say)?

No, the management interface is entirely web-based through your browser.

Also, how difficult is it to run the server headless?

Easy if your motherboard can be booted without a monitor attached, unRAID does not care.  Some MB need a video-card installed to boot, but many MB have the video built-in, so even that is not normally an issue.

  Can I access the GUI with a VLC program?  If no GUI, I assume I just connect to a command prompt via putty?

No GUI, so "putty" is the best way to get to it.

  Will I need a monitor to initially install unRAID, or can I install the OS and configure it all from a putty connection?  (Some of these questions are repeats, but they seem to fit well here too)

You will need a monitor to set the BIOS options to boot from the USB flash drive, and to power itself back on when power is restored... Once those are set, the monitor can be disconnected.

Can I run a media manager on my media center PC that I can use to access and sort files on the share based upon independently entered characteristics (i.e. by type, extension, encoding, quality, resolution, genre, etc.)?  For example, is unRAID compatible with MainLobby, Media Portal, or other programs running on a remote PC?  I assume I just point the media center to the remotely hosted files?  Can I set a MAC address restriction on the media center PC via the unRAID configuration interface that would let the media center access these files without a password (but make other computers still use a user access password)?

No MAC address restriction.. but you can on the Pro version set different "Users" with different permissions and access, including no password access for some user IDs.

What exactly am I paying for?  Is unRAID a modification of the linux kernel, or a closed-source proogram than can be run on Slackware, or what?  What exactly am I buying and why (considering that linux is open-source, with a very open license)?  I don’t think I’ve ever been charged for a program that ran on a *nix distribution before, so was just wondering…

The unRAID "md" drive is GPL source, and its source code is in the /usr/src folder in the file-system when you boot.  You are paying for two proprietary things, the Management interface with its ability to allow you to easily add drives to your array, and manage users, and permissions, and space allocation... and a  proprietary  file-system driver that presents to you a set of "User Shares" representing the combined contents of your physical disk's folders in a single logical share.

What is the fewest number of drives that can be used to build an unRAID array?  2?  (1 drive and 1 parity) (note: I understand that a two drive system is rather pointless, but it would be a good place to start expanding from as I fill up drives)  Will unRAID refuse to work with fewer than 3?

Believe it or not, 1 drive... Yes, it will not be protected from failure, but it will serve up a single data drive.  You need 2 drives for parity protection, 1 data, and 1 parity.

 

Is there any auto-update utility or any way to update the unRAID version or Slackware version without having to rebuild the array, redownload packages, rewrite scripts, etc. (Do I just copy the new files over the old ones - are the config files stored separately?)  How does upgrading Slackware work?  (Again the forums touched on this, but were confusing).

Most times it is simply coping two files to the flash drive and then re-booting.  This most recent release involved a third file, but it was the "memtest86" utility that was also being updated.  No other files are normally involved, and nothing else needs to be "rebuilt"  All the other "config" files are left as they were, and you basically just reboot.  I normally just rename the current "bzroot" and "bzimage" files to  names reflective of their release, then use file-explorer in windows to drag over the new versions of the same two files from the new release to the "\\tower\flash" share, and then reboot.

The forums mentioned that if bad sectors existed on a disk, and a reconstruction was being accomplished due to a failed (other) disk, data could be lost.  Thus, regular parity checks are recommended.  Is parity checking the only way to check for and reassign bad sectors?

No, it is not the only way, but it is a very easy way to read every sector of every disk.  All we are trying to do is read every sector, regardless of if it is in a file-system or not.

  Will Smartctl do this as well?  Reiserfsck?  (This question ties back into the checkdsk equivalent question earlier) - I'm a bit confused as to the differences in preventative maintainance performed by smartctl, parity check, and reiserfsk (there seems to be a bit of overlap, in the area of bad sectors in specific), and which you need to run and how often?

The SMART firmware on many disks can be instructed to run a "long" self-test by the smartctl program.  It can then be used to read the SMART performance statistics.  It will not "fix" any data, whereas running the "parity check" process will.  Best to think of smartctl as a reporting tool only.

 

Tying into the previous question, how do you schedule tests?  Is there a GUI that comes with unRAID or is the current option (still) to write cron scripts for each test you want to schedule?  Does unRAID come with cron and documentation that explains how to use cron, or should I just copy/paste scripts provided on the forums?

unRAID comes with very little documentation... The web-based wiki manual is probably as complete as you might get.  There is certainly no "cron" based documentation I'm aware of, other than it is exactly the same as any other Linux/unix system in its configuration at the command line level.

A 'cache drive' was mentioned in the forums once, but no details given - what do you mean by using a cache drive to increase performance?  Does this count against the 16 disk limit?  What exactly is it for and how do you set one up if necessary?

A cache drive is one that is NOT part of the 16 drive limit, but IS included in the consolidated "User Share" view of your data.  If configured when you write to the array your data would be written tot he cache drive.  The cache drive is NOT protected by parity.  Because it does not have a parity calculation involved, you can write to it faster.  So it helps when writing to the array as far a speed is concerned.

At a later time, usually during the night, scheduled by "cron", a "mover" script moves the data on the cache drive to one of the drives protected by parity.  That process still takes as long as it would have otherwise, but since you are not involved, the time it takes is not as important.  Once moved from the "cache" drive, it is erased from the cache drive.  If you access the data before it is moved, it will be read from the cache drive.  As far as you can see from the LAN "User Share" the cache drive does not exist, it simple allows faster "writes" to the array, but at the expense of no parity protection until the data is moved to the protected drive.

I have an APC, and would want to have the computer autoshutdown if it had to run on APC power for more than a minute or two.  There is a many-pages long, confusing forum topic on this issue, which seems to something being currently worked on in various Slackware/unRAID versions.  Obviously, the software that came with the APC will not run on slackware.  I read about a possible solution on the forum, but it seems very involved.  Is it still necessary to recompile the slackware kernel (which I have no idea how to do) and install additional drivers to enable APC powerdown support?  How hard is this to set up?

Pretty easy if yours has a USB interface, as there is an install package un the unmenu package manager for it.  If it has a serial I/O you might have to tweak the apcupsd configuration file.  Support for it is already in the kernel, no need to recompile anything.  The software that came with the APC UPS is not used at all.  As you said, once installed, if an extended power outage occurs, the server is shut down gracefully (if you have the powerdown script installed), and then powered down.

I was originally going to email this to the developers, but figured I'd get as much answered as I could on the forums first, then email any additional questions before making a decision.

 

Thanks to all respondents in advance for your help, and I apologize if these questions have been already answered (I've spent many hours browsing the forums - yes, I HAVE read all the FAQS/documentation as well - and some answers I could not find or were confusing).  Most of my questions can likely easily be answered in a few words.  Some of them (such as the APC question) have been answered, but not concisely, or inconsistent answers were given.  In these cases, please do no simply post a forum link, as I have likely already read the topic and am still confused.

Ask specific questions, and you will usually be directed to the wiki, especially the user-contributed half of it.  It is very actively kept up to date.

I have a fairly good knowledge of computer hardware and some programming skills (read: it’s been awhile since I had two write any code).  I have an extensive knowledge of Windows, but little practical linux experience (read: I know what I want to do, and often if it can be done, but not necessarily the programs or commands to actually do it in *nix).  I know how to configure network security, and although I’m pretty sure I can learn how to write my own scripts if I need to, if I would have to spend days writing and debugging scripts to get unRAID to do what I want it to, I might as well go with RAID6, especially as each new script I write, debug, and primarily use myself would be another potential failure point in what I want as a relatively reliable data archive.  I am willing to spend a week or two of spare time to set this array up, but want it to work without much intervention on my part afterward, aside from adding the occasional drive overnight (hence, many of the above questions concern the stability, reliability, accessibility, and automation of unRAID, rather than ‘how it works’).

 

Sorry for the insanely long post, but I think this sums up pretty much all the concerns I have about unRAID.  If someone familiar with the system could take 15 minutes and type up a couple words in answer to each question point by point, it would be much appreciated.

 

Thanks again,

Zithras

 

To be honest, NOTHING will suit everybody out-of-the-box. No NAS, no server, no product.  I choose unRAID because it could be extended, and because it was built on an established base of software.  It did exactly what it was described, it has held my media collection for the past three years... I've had disk failures, I've had cable failures (this past week actually, and I replaced an intermittent "Y" power splitter in my server. ) but through it all, all my data has been safe.

 

This took *way* more than 15 minutes.

Joe L.

Thank you Joe, Bubba, Weebo, and Jared, for your replies.  (Especially Joe and Bubba for taking the [likely large amount of] time for the detailed breakdown replies!) 

Apparently, a single message in this forum cannot be over 20000 characters...

I've been looking at this awhile, and spent most of the day thinking up any relevant questions/problems I might run into, and it really says alot about this community (and how active it is) that someone (two people, in fact) replied to all of my many questions within a day! 

Before browsing the forums, I was a bit worried that unRAID was some guy's solution that he came up with out of his basement, and that would be very buggy and likely go away in a year or so, leaving me stuck with an expensive unusable server in two years, but it certainly seems that unRAID has a strong community, is easy to use, and is here to stay! (And, in the event of unexpected dropping of the project, it seems that the data on an unRAID server would be easily transferred to whatever its eventual replacement may be).

 

I feel much more comfortable with unRAID now.  As far as what I need the data backup for - well - the original plan was to set up a RAID6 server - but unRAID is less than half the cost, and seems more reliable so instead of burning DVD backups, why not make two

(The cost of unRAID, assuming you find the drives on sale, seems to be little more than DVD backups on a per-GB, and doesn't need to be reburned every few years.  Plus, whenever you're storing hundreds of DVD-rs, some will eventually degrade without you noticing - a backup unRAID server has parity checks to help prevent this, and seems like a much better backup solution, as well as a primary storage solution )

As many others have argued, I am firmly in the 'RAID does not replace backups' camp - but two unRAID servers would replace backups for me in this case - the data isn't nearly important enough for offsite storage, but I'll be annoyed enough at having to rerip it that an onsite backup is probably worth it (in a different room, on a different UPS)

 

Just one more (repeated) question (because it's important):

Just to double-check, none of you regular unRAID users here (Weebo, Bubba, Joe, et al) have ever experienced data corruption while transferring files over the network in a properly configured unRAID setup, correct?  (I.e. the built-in tcp-ip error checking over samba has always been sufficient for you)?

 

Much happier now,

Zithras

Just to double-check, none of you regular unRAID users here (Weebo, Bubba, Joe, et al) have ever experienced data corruption while transferring files over the network in a properly configured unRAID setup, correct?  (I.e. the built-in tcp-ip error checking over samba has always been sufficient for you)?

 

In my years of using unRAID I have not had any data corruption due to issues with unRAID, the MD driver or Recovery.

I can second that...since Dec 05 I haven't had a problem with transfers.  To that extent bad disks either (knock on wood)    We did loose a laptop drive that fall with a lot of stuff on it that wasn't backed up unfortunatley...giving me leverage to purchase Unraid with my wife.

 

I should also note I'm currently migrating to an all SATA setup and transferring over 2.5 TB from one to the other in unraid and its going smoothly.

I initially had data corruption issues, but they turned out to be caused by a faulty motherboard.  Since then, I've tested extensively and never found any problems.  Partly because by nature I tend to double and triple check things, more than most, and partly from long ago experiences with data corruption in network transfers, I almost always file compare all copies across the network, just to be sure the saved copy is intact, perfect.  I would have seen a problem if there was one.

 

As far as I can recall, there have been a few cases of data corruption, but all have been traced to either bad motherboards (with early nForce chipsets), or bad memory.

I have never had data corruption issues.  I've only been using unraid for about 9 months now (not near as long as the 'hero members' in this thread), but I have no reason to ever expect that I will start having such issues.  One thing I have experienced is failed transfers when sending huge files.  But, there is a solution to this.  I found that when I am moving large files (i.e. DVD ISO's) to always write to a drive share and not to a user share.  Reading from user shares has added advantages, but when writing, write to a drive share.  It's more direct and much more stable.  Also, that way you have much more control over which drives your files are saved to.

I've never had corruption, and my largest type of usage would show it -- my EnCase images some upwards of 100GB, all have MD5 checksums, and I have to recalc and compare them at several stages of analysis.  Never had a singly error.

BubbaQ and Joe L., you deserve a commision on this one. I can't believe the number of questions!!  It is a pretty good braindump of the forums for the past several months.

 

Unraid sits on top of Linux networking. Zero corruption on supported hardware. I've run on two different MBs and moved many terabytes of data back and forth. Never a problem.

 

It was never said that the big advantage of RAID over unRaid is write performance. UnRaid is very slow compared to RAID. I'd never want my SQL DB on an unRaid array. But for media files it works great. The cache disk feature moves lengthy transfers to the middle of the night giving APARENTly faster writes.

 

A huge advantage of unRaid is its ability to use different sized disks and incrementally add storage space. No RAID solution provides this capability.

So noted.  I'll run manual .sfv checks on the first 20-25 files transferred, and if they work, I should be okay.

 

Zithras