AVS › AVS Forum › Video Components › Home Theater Computers › SnapRAID: An Open Source alternative to UnRAID and FlexRAID
New Posts  All Forums:Forum Nav:

SnapRAID: An Open Source alternative to UnRAID and FlexRAID - Page 3

post #61 of 213
Quote:
Originally Posted by Defcon View Post

There is no inherent disadvantage of a GUI, quite the contrary in fact. I can understand you focusing on cmd line though.

I understand, and I also agree.

The point is that SnapRAID is only a tool in the many. If you had one GUI for SnapRAID, one for smartctl, one for hdparm, one for dmcrypt, and so on, it will be worse. At least command lines are similar interfaces, that can be easily used together in a scripting language.

That's really needed is ONE integrated and unified GUI for all these tools, taking care of all the aspects of managing a disk array. I would also use it
post #62 of 213
This time I exclude "Application Data" folder as in the config file below but I still have the same error. I don't get iTune error anymore. It is weird.

parity L:\\par\\parity
content L:\\par\\content
disk d1 C:\\Users\\mkz71\\
disk d2 E:\\
disk d3 H:\\
disk d4 K:\\
disk d5 U:\\
exclude *.bak
exclude tmp\\
exclude C:\\Users\\mkz71\\AppData\\Local\\Application Data\\
block_size 256

Quote:
Originally Posted by mkz71 View Post

It seems that the error keeps changing.
I tried this time and it shows different error.

C:\\snapraid-1.0-windows-x86>snapraid sync
Self test...
Scanning disk d1...
Error in stat file 'C:\\Users\\mkz71\\/AppData/Local/Application Data'
post #63 of 213
Quote:
Originally Posted by mkz71 View Post

disk d1 C:\\Users\\mkz71\\

I would not include C:\\Users\\mkz71\\ folder at all. It won't work.
SnapRaid is by definition a snapshot RAID, i.e. good only for static data.
Data in user folder is dynamic, programs lock files, constantly change them, invalidating parity.

The usage pattern is:
- determine data chunks that you rarely change, i.e. all media files and backup copy of documents.
- create parity
- update parity after you manually add/remove files in your data folders.

User folder breaks it because it will be updated without your control.
post #64 of 213
I'm having a similar problem as mkz71. My media files are in the root of the drive, but unfortunately that means that the System Volume Information folder needs to be excluded since it's also in the root of the drive. I've tried about every variation I can think of to exclude that folder, but I keep getting the following:

Self test...
Scanning disk d1...
Error in stat file 'N:\\/System Volume Information'

Anyone know the syntax? Do I need to give up and stick the media files in a folder in order to work around this issue?
post #65 of 213
Quote:
Originally Posted by ramdac View Post

I'm having a similar problem as mkz71. My media files are in the root of the drive, but unfortunately that means that the System Volume Information folder needs to be excluded since it's also in the root of the drive.

Unfortunately it isn't possible to always exclude special directories and files like System Volume Information.
The problem is that in this case the error is got even before the filter can exclude the path.
You can exclude the directory that contains them, but obviously, not in the root case.

Anyway, it's a thing I'm going to fix in the upcoming 1.1 version. Few days to wait.
post #66 of 213
Quote:
Originally Posted by amadvance View Post

Anyway, it's a thing I'm going to fix in the upcoming 1.1 version. Few days to wait.

Thanks for that explanation. I'm looking forward to the next version.
post #67 of 213
Quote:
Originally Posted by ramdac View Post
I'm looking forward to the next version.
SnapRAID v1.1 released at http://snapraid.sourceforge.net/

Thanks to all the AVS Forum for the support and the help!
post #68 of 213
Hey !

I've created an account especially to thank you Andrea for your work.

I've used Snapraid V1.0 on a Windows7 64bits (NAS+HTPC+torrent dl) for about a month and totally love it, so I'm pleased to see this new release 1.1 with a few new minor features and performance increase on the Windows side. Looks great !

I've just download it, reconfigured the conf, and started a new sync form scratch.

Configuration:
_ AMD Zacate APU E-350 1.6GHz (low power CPU!) on a Sapphire Pure Fusion Mini E350 motherboard.
_ 4Go DDR3
_ Windows 7 64-Bits on a Vertex SSD
_ 5 WD 2To Disk (WD20EARS) for Data
_ 1 WD 2To Disk (WD20EARS) for Parity

All disks are connected using SATA3 (even if not SATA3 ready of course), no USB or anything else.

My 2To disks are NTFS formated with 64K block-size and WesternDigital Advanced Format is activated. 95% of my data total volume is MKV files with an average size of about 7 Go. The other 5% are FLAC music (~20Mo), photos (~3Mo), and small text (but not many files).

Conf file:
Code:
# Example configuration for snapraid for Windows

# Defines the file to use as Parity storage
# It must NOT be in a data disk
# Format: "parity FILE_PATH"
parity P:\\parity

# Defines the file to use as Q-Parity storage
# If specified, it enables a double failures protection like RAID6
# It must NOT be in a data disk
# Format: "q-parity FILE_PATH"
#q-parity F:\\qar\\q-parity

# Defines the file to use as content list
# You can use multiple specification to store more copies of the file
# It's suggested to have at least N+1 copies of the file, where N is the number of parity files.
# It must NOT be in a data disk
# It can be in the disks used for parity storage
# Format: "content FILE_PATH"
content P:\\content
content C:\\SnapRaid\\content

# Defines the data disks to use
# The order is relevant for parity, do not change it
# Format: "disk DISK_NAME DISK_MOUNT_POINT"
disk d1 W:\\
disk d2 X:\\
disk d3 V:\\
disk d4 U:\\
disk d5 T:\\

# Defines files and directories to exclude
# Remember that all the paths are relative at the mount points
# Format: "exclude FILE"
# Format: "exclude DIR\\"
# Format: "exclude \\PATH\\FILE"
# Format: "exclude \\PATH\\DIR\\"
exclude Thumbs.db
exclude \\$RECYCLE.BIN\\
exclude \\System Volume Information\\
exclude \\Diskeeper\\
exclude \\Windows Backups\\
exclude \\Incoming Torrents\\

# Defines the block size in kibi bytes (1024 bytes).
# Default value is 256 -> 256 kibi bytes -> 262144 bytes
# Format: "block_size SIZE_IN_KiB"
block_size 128
I use a 128K block size in SnapRaid, cause I think I have enough RAM to handle it (and it would limit the space lost if I correctly understood)

I currently have about 5To of Data spared non-uniformly at all on the data disks: i.e. 2 data disks are almost empty, while two others are almost full (I keep about 200Go free on each disk). I guess this is not optimal from a performance point of view, but we'll see.

A snapraid -T on my system:
Code:
Compiled with gcc 4.5.1
Speed test with 4 disk and 262144 buffer size...
memset 1812 [MB/s]
MD5 252 [MB/s]
Murmur3 1116 [MB/s]
RAID5 int32x2 1692 [MB/s]
RAID5 mmxx2 1745 [MB/s]
RAID5 sse2x2 2481 [MB/s]
RAID6 int32x2 807 [MB/s]
RAID6 mmxx2 1304 [MB/s]
RAID6 sse2x2 787 [MB/s]

The new sync is running: console shows a steady 205-215 MiB/s for the moment.
I will report some performance benchs during the rebuilt.
Snapraid.exe process uses 1.6Go of RAM and an average of 45% of my little CPU power.


2 quick questions now, according to my detailed configuration:
_ Am I right to use a 128K block size in SnapRaid ? Any benfit to use 64K or 256K ?
_ Would I see a performance increase on the Snapraid syncronisation process by trying to "equalize" (can't find the damn word but I think you'll get the idea ) the data volume accross all my data disks ?

Thanks in advance for your answers, thanks again for your work !
Sorry for my english...


EDIT:
* 7% of the global syncronisation done for the moment: still sitting at 210 MiB/s according to the DOS console, but snapraid.exe RAM consumption has dropped to about 350 Mo.

* 45% done so far: 222MiB/s now, snapraid.exe still uses 45% of my CPU and about 350 Mo RAM.

* 86% : 188 MiB/s. snapraid.exe only takes 30% of my CPU now.

* 100% ! : 170 MiB/s finally and about 8 hours.
post #69 of 213
Oups, I forgot to mention: when I created my first array on the same machine using v1.0.x86, I remember that It was running at an average of 120 MiB/s (but maybe because I was reading data essentially from 2 disks, whereas I am reading data from 3 disks now, no ?)

At this moment, I thought this speed value displayed in the console was the average writing speed value onto the parity disk (120 MB/s is the exact value I get when I benchmark my disks using ATTO with large files, coincidence ?).
Wel, obviously it is not the writing speed... my little 5400RPM green drives definitely can not sustain 220 MB/s writing speed.

So, what is exactly this value ? Is it the total amount of read/write data on all disks (data+parity) each second ?
post #70 of 213
Quote:
Originally Posted by Shaaden View Post

2 quick questions now, according to my detailed configuration:
_ Am I right to use a 128K block size in SnapRaid ? Any benfit to use 64K or 256K ?
_ Would I see a performance increase on the Snapraid syncronisation process by trying to "equalize" (can't find the damn word but I think you'll get the idea ) the data volume accross all my data disks ?

128KB is perfectly fine. In real it was the first choice for default. But I increased it after considering the memory usage. I do not suggest to go under 64KB because it may have negative effect on the disk performance because 64KB is the basic disk cache block of Windows.

Equalizing the fill level of disks you will get a shorter "check" time and only at the first "sync". The total time for these commands is at first approximation the slowest time required to read all the data present in the the most filled disk. If disks have the same filled size, you are minimizing this time.

Sync commands after the first are mostly influenced by the amount of data modified and added, and not by the fill level.

Quote:


* 100% ! : 170 MiB/s finally and about 8 hours.

Congratulations!
post #71 of 213
Quote:
Originally Posted by Shaaden View Post

At this moment, I thought this speed value displayed in the console was the average writing speed value onto the parity disk (120 MB/s is the exact value I get when I benchmark my disks using ATTO with large files, coincidence ?).
Wel, obviously it is not the writing speed... my little 5400RPM green drives definitely can not sustain 220 MB/s writing speed.

So, what is exactly this value ? Is it the total amount of read/write data on all disks (data+parity) each second ?

The speed is the summed read speed of all the data disks. Parity read/write is not counted at all.

For example, if you have two data disks reading at 100 MB/s, you should see 200 MB/s.
Even if in true you are processing 300 MB/s because parity is also writing at the same speed. But keep in mind that all the disks have to operate at the same speed, so the slower is the limiting factor.
Also, if a disk has less filled space, when finished reading from it, it doesn't contribute anymore at the speed computation.
post #72 of 213
I'm not able to create the inital sync. The sync process seems to have a problem with unicode characters. Everytime a special character is used in file names (e.g. asian characters), the following error appears:

N:\\SnapRAID>snapraid.exe sync
Self test...
Loading state from D:\\SnapRAID\\content...
No content file found. Assuming empty.
Scanning disk d1...
Error in stat file/directory 'K:/Audio (A-M)/Dire Straits/[1978] Communique [24-96] [VINYL]/04 - Communique¦.flac'. No such file or directory.

In this case the correct filename is "04 - Communiqué.flac"

Is there a workaround for this problem?
post #73 of 213
Hi!
I just finished building a PC from spare parts mostly for SnapRAID testing. Installed version 1.1.

I can confirm that it doesn't allow to create initial parity with file names in non-English encoding.
My system is Win 7 pro with default English locale. I have lots of files named in different Western and Eastern European languages, no Asian characters.
Below is the error message, I can't proceed further.


Scanning disk d1...
You do not have read permissions for file 'D:/Data/Docs/?????????.txt'.
Fix the permissions or exclude it in the configuration with:
exclude /Docs/?????????.txt
or exclude the whole directory with:
exclude /Docs/


PS. I can see meaningful file names in Windows Explorer and Total Commander. dir in command prompt displays question marks as above though. But I can open the file from command prompt. In the example above typing "type ?????????.txt" in command prompt displays the content of the file.
I think, even if SnapRAID can't read file names correctly it should be able to read the content and use it. FlexRAID works on such files. That's a dealbreaker, sorry.
post #74 of 213
Thanks for the answers !
post #75 of 213
Quote:
Originally Posted by schecke View Post

I'm not able to create the inital sync. The sync process seems to have a problem with unicode characters. Everytime a special character is used in file names (e.g. asian characters)

Surely it's something to fix. In Linux it just works, and in fact I haven't tested this specific case in Windows. Supporting Windows is really an exception over exception... but we are near at the end

Thanks for reporting!
post #76 of 213
SnapRAID v1.2 released at http://snapraid.sourceforge.net/

This is a maintenance release to fix the use of unicode file names in Windows. No change in Linux and other OS.

Ciao!
post #77 of 213
Quote:
Originally Posted by amadvance View Post
SnapRAID v1.2 released at http://snapraid.sourceforge.net/
Thanks a lot for the update. Non-English character bug is fixed.

There is another problem, however. It seems not to like file names starting with space?

c:\\SnapRAID>snapraid.exe sync
Self test...
Loading state from C:\\SnapRAID\\Content\\file.bin...
Not found, trying with another copy...
Loading state from F:\\Contentfile.bin...
No content file found. Assuming empty.
Scanning disk d1...
Unsupported name ' - movie trashorama radio.mp3' in file ' '.
c:\\SnapRAID>

I did have a file named exactly " - movie trashorama radio.mp3" - space, dash, etc. I recorded internet radio with WinAmp + Streamripper, this is how Streamripper named the file. BTW, WinAmp plays it.
After I removed few files like that and ran it again SnapRaid said:

Scanning disk d1...
Unsupported name ' - .mp3' in file '☻'.


It's probably the same issue. At that point I just moved away all recorded mp3s. This allowed SnapRaid to start creating parity.

Currently creating parity at 1%. ETA in 9 hours. Will update the status in the evening.
Thanks again.
post #78 of 213
Hi!
In short, most things work well.

And now I'm sorry for such a long post, but I think details of the test as well as some problems may be important for some.

Initial parity build lasted about as long as it was stated in the beginning - about 9 hours. I/O speed varied from 60 MB/s to 160 MB/s. Obviously because number of disks being read varied from 1 to 3. I believe it should scale to whatever, 300-500+ MB/s and more with more disks if there is no SATA controller or bus bottleneck.

Added 46GB blu-ray ISO. sync time - ~16 minutes.
Removed that ISO. sync time - ~13 minutes. Excellent speed.

Parity overhead is ~0.9GB with 2TB data drive being almost full. OK.
Content file size ~500MB.

Memory usage - 420MB Peak Working Set with 256K block. Very good.
CPU usage - mostly within 13-26% with rare occasional peaks to 30%. That's counting both CPU cores and I noticed SnapRAID is single-threaded?
2 Shaaden - 45% of your CPU usage means SnapRAID is using 90% of one core. It won't use more than 50% of dual core CPU. The CPU is most likely a bottleneck in your case.

Then I deleted a folder with ~10GB of files and restored it with fix -f foldername. Most was restored in ~5 minutes and I compared what was restored bit-to-bit with original - perfect.

Problem: Non-English directory names. These names became garbage and if folder name contains at least one non-English character no files in such folders are restored. All files are gone and there is one unrecoverable error per folder in output.

In process of restoring Matrix.Reloaded.1080p.mkv was not in non-English folder. It was restored bit-to-bit fine. But then why to have thousands of such messages in output?

2501960: Data error for file test/Matrix.Reloaded.1080p.mkv at position 6957
2501960: Fixed data error for file test/Matrix.Reloaded.1080p.mkv at position 6957
2501961: Data error for file test/Matrix.Reloaded.1080p.mkv at position 6958
2501961: Fixed data error for file test/Matrix.Reloaded.1080p.mkv at position 6958
2501962: Data error for file test/Matrix.Reloaded.1080p.mkv at position 6959
2501962: Fixed data error for file test/Matrix.Reloaded.1080p.mkv at position 6959
2501963: Data error for file test/Matrix.Reloaded.1080p.mkv at position 6960
2501963: Fixed data error for file test/Matrix.Reloaded.1080p.mkv at position 6960
2501964: Data error for file test/Matrix.Reloaded.1080p.mkv at position 6961
2501964: Fixed data error for file test/Matrix.Reloaded.1080p.mkv at position 6961


Questions:
Can single-threading become a bottleneck with reasonable fast CPU when working with 12-16 or more HDDs at full SATA controller speed?

Can we have a log file? I anticipate that in some really bad cases we may have thousands of errors and all of them might be needed in order to automatically create problematic file list. Console buffer would be overflown.
Currently errors are printed for each block (I think?) so they will flood output.
I tried to redirect standard output to file but I lost real-time progress. A simple fix would be to direct error report to error stream and leave progress in standard stream. Like stderr vs stdout. That is in at least in Windows, I forgot how to program under Linux.

I once accidentally killed SnapRAID process when I wanted to test small data folder recovery and SnapRAID said it would take 9 hours to finish. Then I recalled that there is -f option for that, so I killed it and tried to recover. I did finally recover but I think it took several iterations of adding the same data folder back, running sync, then deleting the folder, running sync again, adding, syncing, deleting and then fix would work.

Should be only one iteration but I'm not sure in what I did because if you list folder name without ending slash after "-f" SnapRAID will say "Nothing to do. No error". A bit confusing.

Here is what I did and I propose to try the following scenario:

Initial conditions: sync has been done and parity is ready.
1. Delete some small data folder
2. Run -fix.
3. Kill SnapRAID process after it starts printing progress.
4. Try to recover without rebuilding whole parity.

I will definitely re-test this again.
For now I simulated hard drive loss and left it restoring full 2TB of data overnight. I don't have non-English characters on that drive and I will compare it bit-to-bit with original.


PS. Test setup:
I have only 4 spare HDDs that I could use for this build. Two 2TB WD Greeen, one fully filled with data, one for parity. 750GB drive and 320GB system drive with ~200GB partition for SnapRAID data.
CPU is Pentium E6500 @2.93GHz in G41 (ICH7 SATA controller) mobo.
For this test I made a data set of ~87,000 files and ~4600 directories in total on 3 data drives. Ranging from small text files to blu-ray ISOs with names in different languages, etc - the most comprehensive data set I could come up with.
post #79 of 213
Nope. Didn't work.
I wrote that I "left it restoring" but I didn't, I assumed I would do.
I found some non-English file/folder names on that drive and decided to clean it for pure experiment. Deleted 70GB, copied 60GB of legit data from server to refill and ran sync.
Here is what I got:

Syncing...
Error growing parity file 'F:\\parity\\file' to size 2057372434432. Do you have en
ough space? Input/output error.
WARNING! Without an accessible Parity file, it isn't possible to sync.


My maximum data size even decreased by ~10 GB comparing to the last sync.
It could be a consequence of my previous manipulations, not sure. This is something to test.

Deleted the parity and content and started creating parity from scratch.
post #80 of 213
I re-created parity from scratch so there should be no mistake now.

My 2TB drive had 20GB of free space.
1. Deleted 40GB
2. Added 35GB
3. Ran sync.
Got the same error: Error growing parity file 'F:\\parity\\file' to size 2016715735040. Do you have enough space? Input/output error.

I do have enough space. Does SnapRAID add new files without taking into account deleted files when it calculates projected parity size?
post #81 of 213
Quote:
Originally Posted by micksh View Post

I do have enough space. Does SnapRAID add new files without taking into account deleted files when it calculates projected parity size?

Thanks micksh for the report. You are correct, adding and deleting files at the same time, may require more space than effectively required. This is because files are deleted only after adding the new ones. It's something that can be fixed.

There is also the limitation of not allowing a space as a first char in a file name. But also this one can be easily fixed.

Sorry for the late answers, I was in Germany for work, and I'll try to get in sync with messages in the next days.
post #82 of 213
Quote:
Originally Posted by micksh View Post

In short, most things work well.

Good!

Quote:
I noticed SnapRAID is single-threaded?
Can single-threading become a bottleneck with reasonable fast CPU when working with 12-16 or more HDDs at full SATA controller speed?

Yes. It's single threaded, but it aks the OS to do all the file operations in background, so in fact, there isn't a real need of a multithreads architecture.

The only potential bottleneck could be the hash and raid computations. But if you try with "snapraid -T" you will see speeds of GB/s for these operations. This is the reason why I switched from MD5 to murmur3 hash in version 1.1. In version 1.0, the MD5 speed was a limitation.

Anyway, if in future a potential gain could be obtained, I will switch to multithreads.

Quote:
Problem: Non-English directory names. These names became garbage and if folder name contains at least one non-English character no files in such folders are restored.

Something to check. Another know problem is when you specify such names in the command line. Like using the -f option. Also when these file names are printed on the screen.

Quote:
Can we have a log file? I anticipate that in some really bad cases we may have thousands of errors and all of them might be needed in order to automatically create problematic file list. Console buffer would be overflown.

hmm. What you are describing about stdout/err is already supposed to work. Have you tried with:

snapraid check 2> result.log

(note the "2>")

If should redirect all the pedantic messages to the log file, and leave the progress status on the screen.

Quote:
Should be only one iteration but I'm not sure in what I did because if you list folder name without ending slash after "-f" SnapRAID will say "Nothing to do. No error". A bit confusing.

Without the ending slash, the file name pattern is applied only at files and not to directories. I copied this syntax from the rsync utility. But you are correct. Likely a simpler one would be better. I will reconsider it.
post #83 of 213
SnapRAID v1.3 released at http://snapraid.sourceforge.net/

This is another maintenance release to fix know issues. Mostly things reported by mkz71

Ciao!
post #84 of 213
Thank you
post #85 of 213
"You can start using SnapRAID with already filled disks."
Does this also include stripped disks (two drives in RAID0)?
post #86 of 213
Good work !

Will install it and post feedback next week !
post #87 of 213
Quote:
Originally Posted by Looneybomber View Post

"You can start using SnapRAID with already filled disks."
Does this also include stripped disks (two drives in RAID0)?

Yes, SnapRAID works at filesystem level. It can use any mounted filesystem.

For RAID0, the two disks are seen as a single filesystem, so you will need a parity disk of the same size.
post #88 of 213
Quote:
Originally Posted by amadvance View Post

Yes, SnapRAID works at filesystem level. It can use any mounted filesystem.

For RAID0, the two disks are seen as a single filesystem, so you will need a parity disk of the same size.

That is good news. I will have to try this out when I rebuild my computer.
post #89 of 213
Quote:
Originally Posted by amadvance View Post

snapraid check 2> result.log
(note the "2>")

Yes, that surely works. I forgot about "2>" stuff. Thanks.

Recreated parity with 1.3, did few tests.
- "SnapRAID adds new files without taking into account deleted files when it calculates projected parity size" - This now works. It's fixed.
- "File names starting with space" - Works. Fixed.
- "Restoring files when parent folder has non-English characters in name" - Could not test because of reasons below.

I decided to erase my 2TB drive and restore it. Restored about 10% then SnapRAID stopped with error:
Error opening file 'D:/Data/Downloads/bb data/AUTORUN.INF'. Permission denied.
DANGER! Without a working data disk, it isn't possible to fix errors on it.
Stopping at block 586869

This file has read-only, archive and hidden attributes. SnapRaid didn't complain when it created parity and I didn't change my data.
The file can be opened in notepad just fine.

So it stopped, some files were restored completely but some only partially. I could not find any indication of what was fully restored and what has chunks of good data and chunks of zeroes.

Two questions regarding this:
- Does SnapRAID open original data files for read-write? If it is so, I find it dangerous. There should be no reason to open supposedly read-only data for writing.
- If some small file in original data set is damaged (due to bad HDD sector or accidental deletion) will it prevent me from restoring most of my data?

I thought it would work like this: If some file used in parity creation can't be opened it won't allow to restore only file(s) that depended on that particular parity block(s). This file would be marked as unrecoverable, program would skip it but it would continue trying to recover what it can, based on next valid data it can find.

In other words, one original data file is lost, this invalidates block(s) of parity, this makes impossible to restore files that overlapped with now invalid parity block(s).
This is expected, but I thought everything else that still has matching parity and good original data should be restored.
Is it the ultimate design goal?
post #90 of 213
Hello,

First, I'd like to shout out THANK YOU for starting an open source project around this concept, I look forward to others joining. If I had more skills, I would for sure help dev. If you want help on editing on the web pages for english, lmk.

1) I'll never support the closed-source flex-raid, would rather pay for unraid
2) I hope those applying themselves to flexraid aren't just the dev's friends and see the potential and greatness for an opensource platform
3) Tested snapraid out on win7, and for those scared or timid, it is EXTREMELY easy
4) power management and spindown I did not try, that could be frustrating for such users

SETUP: One computer as a host, used various other computers to mount a drive and map a network drive

RESULTS:
I used very small drives
DRIVE A
8gb
DRIVE B
9gb
DRIVE C
16gb
DRIVE P
9gb
PARITY was ALWAYS drive P, and I started with DATA DRIVES A and B
Put only about 233MB on the 2, yes TWO data drives (the 16gb was used to test the recovering method)
Step A:
snapraid sync... everything cool
Step B:
remove DRIVE A
Step C:
insert DRIVE C
Step D:
snapraid fix, appears okay
Step E:
snapraid check, VERY SLOW OVER NETWORK, speed normal locally

I used 4 different computers:
1 on gigabit lan
1 on gigabit lan
1 on wireless a/b/g
1 on wireless N

HDTUNE and wireless N benchmarks show that transfer speeds should be good, however even on the gigabit lan to gigabit lan transfers
(Drive C installed on a network computer and being accessed from the computer with all of the other drives), the fastest speed came up as 5MiB/s

I did A TON of different tests, and every time I tried a network transfer, the speeds were terribly slow when running "snapraid check" after "snapraid fix," ...(snapraid fix was VISIBLY slower as it scrolled through btw).

Locally, the stuff worked GREAT. However, the results of my limited research suggest that something can be improved with network coding.

I really really wish I knew programming to help out more than reporting results, and hope my testing can help out others and you, the developer.

For the curious...if you're going to have ALL drives on a SINGLE computer LOCALLY, run with this, it worked great.

Personally, I'm very tired LOL, and have to think about what I'm going to do, locally might be enough, but will have to test out on different drives. Might get some tomorrow.

Cheers
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Home Theater Computers
AVS › AVS Forum › Video Components › Home Theater Computers › SnapRAID: An Open Source alternative to UnRAID and FlexRAID