I installed my unRAID server on Thursday, so I am a total newbie to the server should work. But something happened on Friday that made me worry.

I had installed three brand new 500 GB disks, two WD and one Samsung. I had moved data to both of the data disks, connected both PCs and my Sono system to check that the server was working fine. The management console showed everthing ok, temperatures ranging from 24-32 Celcius. Suddenly the Samsung is not accesible and shows up in the panel as unformatted.

 

Since I had not erased the source data yet, I pushed the Format button. (I would like to point out that I couldn't find any other button in the web-interface that made more sense to push.) After format, the disk was empty. No surprise. Next step, I thought was to recover the lost data from parity. After all if the disk had had a hw failure, I would have to reformat the new disk and restore from parity.  So, I initiated the parity check. Which returned parity ok. That was a BIG surprise. I had expected a message telling me that the lost data would be restored based on the parity data. How can it be that I have two data disks and a parity disk up and running, status ok, and after reformatting one of them, parity is ok? Were all the data there, just that I didn't see it?

 

I hope someone can help to explain this, so that my trust in the unRAID server can be restored :-) 

 

I mentioned the unformatted disk in another thread, immediately after it happened, and Joe L gave me the advice to use the reserfsck command. Unfortunately, I had already pushed the format button when I read that.  Also, with a product like this, it should not be necessary to know Linux commands, everything should be managed through the web interface. Basically, I am concerned whether I can delete my source data or not, is the unRAID server mature immature? Or did I do something stupid the way I handled this?

 

Håkon

 

 

I installed my unRAID server on Thursday, so I am a total newbie to the server should work. But something happened on Friday that made me worry.

I had installed three brand new 500 GB disks, two WD and one Samsung. I had moved data to both of the data disks, connected both PCs and my Sono system to check that the server was working fine. The management console showed everthing ok, temperatures ranging from 24-32 Celcius. Suddenly the Samsung is not accessible and shows up in the panel as unformatted.

 

Since I had not erased the source data yet, I pushed the Format button. (I would like to point out that I couldn't find any other button in the web-interface that made more sense to push.) After format, the disk was empty. No surprise. Next step, I thought was to recover the lost data from parity. After all if the disk had had a hw failure, I would have to reformat the new disk and restore from parity.  So, I initiated the parity check. Which returned parity ok. That was a BIG surprise. I had expected a message telling me that the lost data would be restored based on the parity data. How can it be that I have two data disks and a parity disk up and running, status ok, and after reformatting one of them, parity is ok? Were all the data there, just that I didn't see it?

 

I hope someone can help to explain this, so that my trust in the unRAID server can be restored :-) 

 

I mentioned the unformatted disk in another thread, immediately after it happened, and Joe L gave me the advice to use the reserfsck command. Unfortunately, I had already pushed the format button when I read that.  Also, with a product like this, it should not be necessary to know Linux commands, everything should be managed through the web interface. Basically, I am concerned whether I can delete my source data or not, is the unRAID server mature immature? Or did I do something stupid the way I handled this?

 

Håkon

 

 

Basically, you asked that your data be erased.  You would have had the same problem regardless of what RAID product you used.

 

Now, if you had simply shut down and re-started, as I had tried to get you to do, it might have recovered on its own. 

If a disk fails, and parity is used to rebuild it, I don't think the new disk is formatted first, as the process of restoring from parity puts an exact image of the old drive on the new and that copy contains the formatting. 

 

So... copy your data once more onto the unRaid array from your other machine.  Never format an "unformatted" drive unless you know it is empty or don't care about its contents.  If you format a drive, it is effectively erased from parity too. Basically, you "did something stupid" by asking the disk to be formatted.  Next time, use the "Restore" button and not anything marked "format"

 

For now, I am assuming it was a loose cable that caused the Samsung drive to appear unformatted.  If you reset the cables and it happens again, suspect the drive itself as having a problem. Let's hope that is behind you.

 

Assuming all your drives are working, try this exercise before you erase the data from your other drives...

 

Put some files on the Samsung drive

Stop the array...

Go to the devices page and unassign the Samsung drive.  This will simulate its failure.

Start the array.  You will see the Samsung drive is marked as missing on the web-display, but you will still be able to read and write to it.  (The parity drive will supply the missing contents...  )

Now... simulate replacing the failed drive by...

Stop the array once more

Re-assign the Samsung drive

Re-start the array.  It should see the (simulated) replacement drive and start re-construction of its contents from parity. (This re-build will take several hours, but during this time your array will be available and you can get to all your files)

 

 

 

 

 

 

 

 

Copy some files onto

Thanks, that explain a lot - and taught me not to use the Format button again.

 

 

What still puzzles me, is that I did push the Stop button before I formatted, but it didn't stop - it just returned to the same page. I don't remember exactly what the screen said, wait... or some other message. Then, when I pushed the Refresh button again, it had not stopped and said nothing about being in the process to stop. It was actually the same as before I pushed the Stop button, showing the disk being unformatted, and with only the Refresh, Stop and Format buttons available. So my assumption was that the system would refuse to stop until the disk was formatted, which I now understand is fatally wrong, but I still would like to understand why it did not respond to the Stop button.  Also, why wasn't the Restore button visible when the disk status was Unformatted?

 

Håkon

 

Thanks, that explain a lot - and taught me not to use the Format button again.

 

 

What still puzzles me, is that I did push the Stop button before I formatted, but it didn't stop - it just returned to the same page. I don't remember exactly what the screen said, wait... or some other message. Then, when I pushed the Refresh button again, it had not stopped and said nothing about being in the process to stop. It was actually the same as before I pushed the Stop button, showing the disk being unformatted, and with only the Refresh, Stop and Format buttons available. So my assumption was that the system would refuse to stop until the disk was formatted, which I now understand is fatally wrong, but I still would like to understand why it did not respond to the Stop button.  Also, why wasn't the Restore button visible when the disk status was Unformatted?

 

Håkon

 

It sounds as if the web-interface got confused when you asked it to be refreshed while it was stopping the array.

I know that on my array it takes a while for it to stop... primarily because it must first spin up all the drives (most of them in my array are asleep at any given time)  If perhaps it was attempting to learn if a drive was formatted or not while it was still spinning up, it might accidentally think it is unformatted when it is unable to read the expected data.  That might be why the drive appeared in the web-interface as unformatted (just guessing here)

 

If you ever get it confused again, and do not see a restore button, but do see a format button, log in via telnet as "root" and then type

stop

 

(Note... at this point the web-interface will be out of sync... it does not know you stopped the array from the command line)

 

once it says the array is stopped, type

reboot

 

Your array will reboot itself and odds are the web-interface will figure out everything just fine and be back in sync.

 

Joe L.

OK, so I will have to be a bit more patient the next time. Joe, thank you for spending your time to explain so well, I really appreciate that!

 

In the believe that I have now done the most stupid beginners mistakes, I have already sent my order for the Plus license, slaughtered one of my old 250 external Maxtor drives and installed it, ready for the license to arrive... This is the first IDE/ATA drive, so I am a bit exited if I will have issues with that.

 

Håkon