| Author |
Message |
|
|
Brahim wrote:I could implement FlexRAID Live! to be Windows specific, but I am backing away from that until further notice.
That would be great, seeing as unRAID is Linux only...
Chances are you will see FlexRAID NAS before seeing FlexRAID Live!.
Well... I can't help but 
|
 |
|
|
Will you focus on a Live implementation, apart from RAID 6 and RAID? (can't make the symbol  ) engines?
|
|
|
|
 Good!
Remember though, underpromise and overdeliver... now you've set yourself with shoulders against the wall, don't you think? 
|
|
|
|
skirge01 wrote:Lastly (for now, that is), I keep seeing mention of just a single parity drive. I'm having a hard time believing that I could have 30 1TB hard drives with data and still only need a single 1TB drive for parity.
There are I/O and performance limits to the number of drives you could realistically calculate parity for. But yes, only one parity drive is needed.
A *really* uneducated example:
Try writing an odd number of columns (your hard drives) and filling them with 0s and 1s for several rows. Now look at them and sum them up, saying that if there are more 0s the result is even and if there are more 1s the result is odd. Write that result in a new column (your parity drive).
Now imagine of taking out any single column (one hard drive failure). By looking at the parity column you can easily guess the value of the missing column for every row.
Obviously XOR operations don't work this way, so the odd number of columns (hard drives) is just to make the example work, you can have an even or odd number of hard drives, it makes no difference. But I believe the principle is close to this.
That's oversimplified, but I believe it can help in explaining parity (well, that's how I understand it, so it might as well be wrong. If it is, I'm deeply sorry ).
|
|
|
|
Well, the RAID is on six drives (one for parity). All SATA. Two 1.5TB drives (one for parity), one 1TB drive, 1 750GB drive and two 500GB drives.
I'm using the onboard controllers on my GA-P35-DS4 motherboard for a total of seven drives (I leave C:\ out of FlexRAID).
XP SP3, 2GB RAM, Q6600.
Just finished an rsych with "INFO : Total process size = 10879963812", it took around 13 minutes. It was a single file of around 10GB added to the array.
My flxr.meta file is over 5MB in size, if that can give you an idea of the amount of files involved. The size of the parity data on the parity drive is 896GB.
Also, is it normal that during the rsynch process really small 00xxxx.flxr files are created? When I originally created the array it respected the 1GB split size, apart from the last file created that was slightly smaller. Now the first two rsynch I have done have produce another two .flxr files, 35KB and 40KB.
Two other questions:
1) Is it possible to run the Host service as a real service, at startup. Running it manually means having a window permanently on the taskbar.
2) I noticed that, once finished the rsynch procedure, FLEXRAID.EXE in task manager is taking 230MB of RAM, while doing nothing. A bug? Shouldn't it release the memory once finished?
|
|
|
|
Good news. I hope you will explain the differences between the standard validate and the quick one.
I tried rsynch with a new 1.3 GB file. It took around 4-5 minutes.
|
|
|
|
Is rsynch faster than validate, then?
I'm worried about trying it and being unable to use the machine for a long time
|
|
|
|
Brahim wrote:
bigbud120 wrote:Does that mean that we should both Validate and Verify after an Rsynch or just do one? I've been doing both. 
Only one.
"Validate" is meant to be faster.
"Verify" is more of a bullet proof verification process while "validate" is a smart/efficient way of verifying the data.
The difference in speed in even more pronounced in RAID6 and RAID∞ configurations.
Brahim, one question on validate. You mention it's supposed to be faster.
Is it "normal" for the validate task, on an array that took around nine hours to create, to take an estimate of three hours?
Also, while the processor occupation is at around 20-23%, the system becomes unbearably slow while running the task (I'm on a Q6600).
Three hours like this makes FlexRAID a non-solution for me, as media files are added on an almost daily basis (and I'm afraid to see how long the synch process could take).
|
|
|
|
Ok, with the new build it appears to be working. The previous error has disappeared.
Thanks!
|
|
|
|
I'm using Antivir for antivirus.
I'll check the new build and report back.
|
|
|
|
|
Nope. H was empty, that's why I tried switching parity to N, which had plenty of space for the test but it sure wasn't empty.
|
|
|
|
|
Both H and N are perfectly functional drives, with plenty of space for the test RAID I was trying to generate.
|
|
|
|
Are you referring to 009?
And is this something to be used on schedulers, right?
|
|
|
|
Nope, no security.
It's a new drive, properly formatted, as far as I can tell. Just to be sure I copied some files in and out of it, with no problems.
|
|
|
|
|
So far it's 100% of the times. I think I have tried at least 10-15 times, with different folders as data point and a whole 1.5TB empty hard drive as parity.
|
|
|
|
Brahim wrote:Ouch!
I guess the issue was more severe than I thought.
What OS are you using?
Very few users encounter this bug, and when they do, it is not often.
In all cases, you will have access to build 009 tonight.
I am under XP SP2, with all security patches installed.
|
|
|
|
Brahim wrote:@gorman,
I am surprised that retrying the task still fails.
It looks like you might need to restart the host application to clear out the bug.
This would be the solution until build 009 is out.
I'll try again, but I already restarted the host several times... 
Edit: I confirm. Tried once more, with different folders to create parity for, nothing changed.
[2008-11-23 10:39:08,109] ERROR: [create] For input string: "_______"
java.lang.NumberFormatException: For input string: "_______"
at java.lang.Void.<unknown>(Unknown Source)
at java.lang.Void.<unknown>(Unknown Source)
at java.lang.Void.<unknown>(Unknown Source)
at java.lang.Void.<unknown>(Unknown Source)
Edit 2: just to be sure, I tried with the command line client, same error (I was using WebUI previously).
|
|
|
|
Brahim wrote:
Both the ArrayIndexOutOfBoundsException and NumberFormatException bugs have been resolved for build 009.
When is 009 coming? I'm getting the same error...
|
|
|
|
You ought to update the homepage. And give, somewhere, the location to download the files.
On the homepage you forward people to a private section of the forum, available to registered users only, in that section there still is no link to download the software.
For any but the most dedicated would-be users this could be a show stopper.
Just saying...
|
|
|
|
|
Link in the private release forum, open to registered users.
|
|
|
|
![[Logo]](/forums/templates/default/images/logo.jpg){kind=logo}
![[Search]](/forums/templates/default/images/icon_mini_search.gif){kind=icon}
![[Recent Topics]](/forums/templates/default/images/icon_mini_recentTopics.gif){kind=icon}
![[Members]](/forums/templates/default/images/icon_mini_members.gif){kind=icon}
![[Groups]](/forums/templates/default/images/icon_mini_groups.gif){kind=icon}
![[Register]](/forums/templates/default/images/icon_mini_register.gif){kind=icon}
Register![[Login]](/forums/templates/default/images/icon_mini_login.gif){kind=icon}
Login