I just put in 3 1TB hard drives into my core.
I went in to the admin->advanced->RAID
I added a new raid array (raid 5)
I added a 3 drives (1TB each) and made sure none of them had a check mark under "as spare disk" (i have done this twice now)
But what happens is that 2 drives become active, and one becomes spare.. (I don't want a spare, I want a 3-disk raid5 array.. In fact, the minimum for a RAID 5 array is 3 drives, so I don't know how or why it is working with 2 active drives)
Also, the size come out to be 1863.03 GB, shouldn't it come out to be 2000.00 GB??
Is this a bug, or am I missing something?
Also, will LMCE alert me if a drive fails so I can rebuild the array with a new one?
you are definitely missing something. when you use 3 drives you only get the space of 2 (3rd drive isnt a hot spare but all three lose 1/3 for parity). look up raid 5 at wikipedia.com. also...it wont equal 2000 because of the fact some jackhole figured 1024 is better than 1000. so binary doesnt exactly equate to what the computer sees. have a look at those and youll understand why it is the way it is. ill see if i can find some links to explain it better.
(oh...and by all means the jackhole was right....it just screws everyone up ;) its because the computer uses binary but people dont think that way. the link i added explains it better than i can)
Space Issue
http://www.md4pc.com/questions/58.htm (http://www.md4pc.com/questions/58.htm)
Raid 5
http://en.wikipedia.org/wiki/Raid_5#RAID_5 (http://en.wikipedia.org/wiki/Raid_5#RAID_5)
I know raid5 only uses n-1 of n disks for space..
Again, the problem is that for this to work you need a minimum of 3 disks... Only 2 are listed as active and 1 as spare. Shouldn't all 3 be active in order for the striped parity to work? I don't want a spare - if a disk on the array fails, I will go get a new one.
(Maybe i'm misunderstanding the meaning of spare in the admin - but on most RAID systems, all 3 disks are active in a 3-disk raid5 array (and yes, I know I will lose space of aproximately one disk due to the parity bytes). I'm just confused as to why it is listed as spare in the admin is all.
Also, it has been several hours since building the array, the hard drive light is still on, and in the admin the array is still listed as "Degraded/Rebuilding" and the "spare" (as the admin is calling it) as Activating..
Should it take this long - considering it is a new array and there is no data to rebuild...?
Actually, looking through the admin, I think I see what is going on..
It appears that the way LMCE creates the array is just confusing.. It is rebuilding the array.. It must just format 2 of the disks (which will tell the array that one drive has failed) and to add the 3rd drive it is rebuilding it into the array...
Very odd, but I think I see how its working now. I've just never seen it done this way..
sorry i was in a hurry. i reread where you said its still calling it a spare. ive never seen it do it like that either. ive only ever used hardware raid also though. you are using the linuxmce software raid right? so all three drives show up as active now right? it does seem weird it should take much time to rebuild with nothing on there.
did the space thing help?
Still rebuilding... About 4 hours into it... Admin shows it as 40%. I'm guessing after its done rebulding, all 3 will show as active - but looks like it will take the rest of the day to figure out.
Yes, it is using the Raid in LMCE. I've never seen it like this either, but I haven't used a software RAID before either. Either its a bug, or it was decided to be the default behavior for LMCE. Would be nice to have an official word on this.
Thanks for the links. I do understand why computers use 1024 instead of 1000 (it is a square of 8), but I was suprised that I could lose 200 GB of space just to partition type overhead. I was really expecting to see much closer to 2000GB - oh well, you learn somethign new every day.
I will report back when it is done "rebuilding" my (not) damaged (and new) array to let you know if it all comes out well in the end.
thanks again
I did the same thing a week ago with 3x750GB disks, and YES, it does mark the last drive as a spare while rebuilding (And that took a day).
If you try rebooting during this process it will restart... So let it finish.
It is actually usable while rebuilding, but then rebuilding just takes a lot longer time...
I'm quite satisfied with the Raid5 space on LinuxMCE. I have a Thecus with 3 disks also and that uses 100GB more space than linuxmce. (it created a 100GB swap partition :(
thanks for the info.. Strange that it behaves this way... At minimum, it should be changed so the user is alerted that the array is building for the first time or something similar.
It has a progress bar in the beginning for formatting, but that only took 10 minutes or so. I had initially figured that it was finished after that step.
i started thinking maybe it should be more than 1800...i found this....and it says 1800 is right....sure is a lot when you think about how little drives used to be.
http://tomorrowtimes.blogspot.com/2007/08/hard-drive-capacity-calculator.html (http://tomorrowtimes.blogspot.com/2007/08/hard-drive-capacity-calculator.html)
One more question about the LMCE RAID...
When I set it up, i had to option for the parent to be either "Core" or "Core/Hybrid"
I chose "Core", but, I wonder what the diference would have been choosing the other?
Just to follow up - After about a 7 hour "rebuild" process on the raid array , all 3 drives so show as "Active" now like they should be.
I still wonder if the implementation in the admin is flawed - these were new drives in a new array, so there should have been nothing to rebuild. I did this 2 times with the same results.