Areca ARC-1220 Loses Drive

S

stncttr908

Limp Gawd
Joined
Mar 27, 2003
Messages
227
Hi all,

I have 5x1TB Samsungs on an Areca ARC-1220 on my file server. I put together this config around 3 months ago and it has been great up until this point. While listening to some music earlier I had an alarm go off on the card. I noticed the first drive connected to the controller had dropped from the array. Alright, I figured, let me reboot and get into the card's BIOS.

Inside the BIOS, the card now thinks I have two RAID5 arrays (??). The first "array" is the first drive, while the second is the other 4 remaining drives. Of course, this first "array" states it's "incomplete" while the second is "degraded" I guess from missing the other drive.

So, you might be figuring "just delete the first one and add it back to the other one" but here's the thing - it works just fine under Linux. I can read/access all my files. From my understanding on RAID5 this shouldn't be possible, should it? Since data and parity bits are spread across all drives, wouldn't the data not be readable sequentially at some point?

Anyway, what do you think I should do? My music collection is the most important thing on the array, and it's being backed up to a 500GB external as I write this. After that, the rest of the data is basically expendable. I'd rather not lose it, but if I were to it's not the end of the world.

Should I just remove the first "array" and try to re-add it to the real array? Thanks for your time and any advice you can lend!
 
if reboot/power off didn't fix it then delete the incomplete array, set the now free drive as hot spare and let the degraded array rebuild itself.
 
axan said:
if reboot/power off didn't fix it then delete the incomplete array, set the now free drive as hot spare and let the degraded array rebuild itself.
Click to expand...
This is probably what I'll end up doing.

I wish I knew about the Samsung thing before I dropped hundreds on this. Maybe I'll sell these Samsungs and grab some Seagates.
 
There is a compatibility issue with samsung drives and raid controllers.

Also, yes, your array will still work, with raid 5 you can lose ONE drive and you can still be fully operational, this is why you can still access everything. Now if another drive dropped out of your array (likley given the compatibility issue), then you will lose everything.
 
Thanks for the replies. If I lost the one drive, would that account for any input/output errors on certain files on the remaining drives, or is that another issue entirely?

Also, to migrate over once I get different brand drives, I'm assuming the best idea would be the rebuild the array, attach three of the new drives and create an array with them, copy the data over (it's < 2TB, so it should work?), and then add the remaining two new drives to the array.

And Ockie, I'm hoping it can hold on for the data copy! I've had this setup three months and this is the first instance of this happening.
 
errors on certain files is another issue.

Yes best way to move to different drives is to setup another array and copy over the data. You could also replace the drives in your current array with new ones 1 at a time and let it rebuild after each.
 
axan said:
errors on certain files is another issue.

Yes best way to move to different drives is to setup another array and copy over the data. You could also replace the drives in your current array with new ones 1 at a time and let it rebuild after each.
Click to expand...


Personally I would create an array with new drives and then transfer from array to array... it would be too scary to upgrade or swap drives in an already degrated array with the risk of losing a drive while building a new drive.
 
swapping drives 1 at a time would be my last resort and I would only use it If I had now way of creating 2nd array and coping the data. Doing 2 side by side arrays migration is a lot safer and faster.
 
Gracias good sirs.

Drive recommendations? I know there's a wealth of info in this section regarding this, but please forgive my anxiety lol...
 
if you want to use cheaper desktop drives then seagate 7200.11
enterprise drives wd re green or seagate es.2
 
ZZF has the 1TB Seagates (which I should have gone with to begin with given my affinity for Seagate) for $129.99/each. I'll probably do it.
 
You are asking for trouble using AS drives in an array. If you value the data, use NS drives. If you insist to use AS drives, then I highly suggest that you take backups at regular intervals (which you should do anyway) as well as run consistency checks since the AS drives are 10 times more likely to experience unrecoverable bit errors, which causes errors in large arrays to appear much more frequently. The consistency checks will uncover them quickly so that the data can be kept "clean."
 
I see. The backups would be regular. How often would you recommend consistency checks? Monthly? Weekly?

Again, great info. Thanks all.
 
madcoder said:
You are asking for trouble using AS drives in an array. If you value the data, use NS drives. If you insist to use AS drives, then I highly suggest that you take backups at regular intervals (which you should do anyway) as well as run consistency checks since the AS drives are 10 times more likely to experience unrecoverable bit errors, which causes errors in large arrays to appear much more frequently. The consistency checks will uncover them quickly so that the data can be kept "clean."
Click to expand...

Quite a few of us with large storage arrays uses AS drives without any problems. The only case I heard of a AS issue was with the new areca sas controller (1680)
 
I would say weekly consistency checks are adequate. There may be other opinions though.
 
Ockie said:
Quite a few of us with large storage arrays uses AS drives without any problems. The only case I heard of a AS issue was with the new areca sas controller (1680)
Click to expand...

Hi Ockie, yes I am one of those as well, I just don't post so much about what I have set up. I have many, many NS drives, and some AS drives. My point is only that there are so many of us doing this, that it may give the impression that such a set up is secure, when it is not at all. The fact that it is working "without any problems" doesn't change the statistics that there can be a failure. The AS drives have a MUCH higher unrecoverable error rate than the NS drives. With AS drives of a terabyte or more each, in very large arrays (around 10 or more), there is a very good statistical chance of errors, even with RAID6. If you run one of these 24 port card, full of AS drives, and run consistency checks often, you are very likely to notice that there are in fact commonly errors that get corrected. The math can't be disputed on the the calculations for these possibilities. I run AS drives as well on some arrays that are not critical for me, and so I am not trying to rain on anyone's parade, just inform those that might not realize the facts around these statistics.
 
As long as I do regular backups of my music it would be REALLY hard for me to justify $400 extra for the NS class drives, but it is definitely something to think about.
 
Thanks for that jbrukardt.

My restore went perfectly in around 5 1/2 hours. My data is all intact and the corrupted mp3 files are fine! Thanks to everyone. I'm still going to swap out the drives for Seagates.
 
stncttr908 said:
Hi all,

I have 5x1TB Samsungs on an Areca ARC-1220 on my file server. I put together this config around 3 months ago and it has been great up until this point. While listening to some music earlier I had an alarm go off on the card. I noticed the first drive connected to the controller had dropped from the array. Alright, I figured, let me reboot and get into the card's BIOS.

Inside the BIOS, the card now thinks I have two RAID5 arrays (??). The first "array" is the first drive, while the second is the other 4 remaining drives. Of course, this first "array" states it's "incomplete" while the second is "degraded" I guess from missing the other drive.

So, you might be figuring "just delete the first one and add it back to the other one" but here's the thing - it works just fine under Linux. I can read/access all my files. From my understanding on RAID5 this shouldn't be possible, should it? Since data and parity bits are spread across all drives, wouldn't the data not be readable sequentially at some point?

Anyway, what do you think I should do? My music collection is the most important thing on the array, and it's being backed up to a 500GB external as I write this. After that, the rest of the data is basically expendable. I'd rather not lose it, but if I were to it's not the end of the world.

Should I just remove the first "array" and try to re-add it to the real array? Thanks for your time and any advice you can lend!
Click to expand...

That's because you CAN read from a degraded array. Do as axan (and yourself) suggested. Let it rebuild, then swap out the drives if they've got compatability issues!
 
You must log in or register to reply here.
Back
Top