April 18, 201016 yr Hello, Like a thread from a few days ago, I also lost all of my data. Nearly new system, latest stable release, and after a couple of weeks running I added a 2nd data drive. After rebooting, I did something wrong and it started reformatting my 1st data drive. It isn't the end of the world, as most of the data consisted of DVD rips, so it's mostly time and effort lost, not irreplaceable data. However, eventually it will hold valuable irreplaceable stuff, and I can't make this mistake again. So, I captured a syslog, but the log appears to contain data only from today, April 18th....but all the 'damage' I did happened yesterday, the 17th. Is there a way to get syslog data from the previous date? Ideally, I'd like to figure out where I went wrong, so that I don't do it again. Thanks.
April 18, 201016 yr too hasty, you picked format before your data drive was mounted probably not re syslog unles s you saved it yesterday.
April 18, 201016 yr Author but if I inadvertently reformatted 'disk1', can't it be restored with data from the parity drive? Isn't this what it's supposed to be able to accomplish if the data hadn't been reformatted, but instead had been lost due to drive failure, and subsequent drive replacement with a new 'disk1'?
April 18, 201016 yr To answer your questions: No. Formatting a drive is no different than writing data to a drive. When you write data to a drive, the Parity drive is updated. During your format of the data drive the system synced Parity which means the Parity drive reflects the changes (the format) of your data drive. If you can, stop the array and keep it offline. Do not write anything to it. There are some steps you can take to attempt to restore your data. There's a reiserfs check rescan option that may be useful in your situation. I don't know the exact details, but the more senior members do. Hopefully Joe L will reply with the steps needed.
April 19, 201016 yr If you have not rebooted, the older syslog contents are in /var/log/syslog.1, or /var/log/syslog.2 If you have rebooted, the older contents are gone as /var/log is in an in-RAM file-system. Second, remove the term "restore" from your vocabulary. You will see a button with that label, DO NOT PRESS IT if you are trying to rebuild a failed disk. It has absolutely nothing to do with recovering data. In fact, it is a "Set Initial Disk Configuration" button. Pressing it immediately invalidates parity, something you will NOT want to happen if you have a failed disk. Third, as explained, formatting a data disk writes the corresponding "bits" to the parity drive. If the data drive were to fail right now, you could rebuild the replacement disk to where it is just freshly formatted. As far as recovering data. You can try this if you have not written to the disk. (Writing to it re-uses the file data blocks, and overwirtes the old files) The basic procedure is outlined here: http://lime-technology.com/forum/index.php?topic=6008.msg57379#msg57379 The parity drive protects you from a failed drive. It does not protect you from deleting all the files accidentally. If you do, the parity disk will allow you to restore to a disk that has all the files deleted. The parity disk will not let you get back an old copy of a file before you modified it. It lets you recover from a failed drive. (one that cannot be read or written) An unRAID array does not replace backups of critical files. See this post for what I am talking about: http://lime-technology.com/forum/index.php?topic=2601.msg21033#msg21033 Joe L.
April 19, 201016 yr Author Thanks to all for the information...I'm trying the reiserfsck -–scan-whole-partition -–rebuild-tree /dev/md1 command now....crossing my fingers and hoping for the best! As for the page about Top 10 ways to lost data....I'm reading it carefully. And I'm revisiting my strategy....I think I'll hold onto my desktop USB drive for backing up my MacBook. Think I'll keep the unRAID just for the DVD's. Thanks again, this information has been a big help to me.
April 19, 201016 yr This has me concerned. How does a pior data disk get formatted like this. There's gotta be a way to protect people from themselves. I've long felt that the single format button on the first page is a problem waiting to occur. I really think it should be an individual drive option. Perhaps in 5.0 this sort of stuff will get flushed out. I hope the scan-whole-disk option works. This is one of the key reasons I want to do a md5db of the files in each disk. Who knows if anything restored is not restored incorrectly.
April 19, 201016 yr This has me concerned. How does a pior data disk get formatted like this. There's gotta be a way to protect people from themselves. The 4.5.3 release has a significant bug where if a disk takes its time replaying journal transactions, or is slow to spin up, it will NOT be mounted when you first reboot/power-up the array. A disk that is not mounted shows as un-formatted, even if it actually has a file-system. This is a error in interpreting the disk not being mounted. The only reason it should show unformatted is if it had no file-system. Instead, it shows unformatted if: - The disk is un-partitioned - The disk is un-formatted - The disk does not contain a single reiserfs partition starting on sector 63 extending to the end of the disk. - The disk contains any other kind of file system (NTFS, ext, FAT) - The reiserfs on the disk is corrupted in some way and refuses to mount. - The disk is replaying journal transactions and the mount has taken longer than the alloted time. Unfortunately, as you described, if any disk fails to mount as a reiserfs, it is shown as un-formatted, and the format button when pressed will re-format it. It is easy to lose data when you yourself asked for the disk to be re-formatted. I've long felt that the single format button on the first page is a problem waiting to occur. I really think it should be an individual drive option. I agree. Perhaps in 5.0 this sort of stuff will get flushed out. We'll see. Tom really needs to track which disks are already in the array. The "Format" should never be applied as it is today. I hope the scan-whole-disk option works. Me too. This is one of the key reasons I want to do a md5db of the files in each disk. Who knows if anything restored is not restored incorrectly. Exactly.
April 20, 201016 yr Author Okee doke....reiserfsck has done many hours of work, and completed 4 passes....lots of data moved here and there. Did the mount disk command, then the start Samba command, but now the disk still shows as unformatted on the unRAID page. I'm guessing that if it's still showing unformatted, that's bad
April 20, 201016 yr Author Sorry about that, my post was premature......a refresh of my unRAID web page shows it to be formatted...sweet! The final result was that about half of the data was recovered....unfortunately for me the it was the other half that I was most interested in. But hey, I learned a whole lot through this excercise, thanks to all of you. I'm going to try not to ever get myself in this situation again.
April 20, 201016 yr 1. The FORMAT button should never have formatted a disk that's already been formatted. 2. You lost data with unRAID. Until recently they used to make it a big point how nobody ever lost data with unRAID. 1+2: You may want to send a little thank you note to Limetech! [email protected]
April 20, 201016 yr Sorry about that, my post was premature......a refresh of my unRAID web page shows it to be formatted...sweet! The final result was that about half of the data was recovered....unfortunately for me the it was the other half that I was most interested in. But hey, I learned a whole lot through this excercise, thanks to all of you. I'm going to try not to ever get myself in this situation again. Did you look in the lost+found directory the reiserfsck process creates on the disk... The files you are looking for might have weird names, but they might be there.
April 20, 201016 yr Geeze, there are at least 4 or 5 active threads here right now where 4.5.3 has caused data loss via formatting good disks. It's about time Limetech pulls that release and also gets a new fixed version released to stop this mess. Peter
April 20, 201016 yr Geeze, there are at least 4 or 5 active threads here right now where 4.5.3 has caused data loss via formatting good disks. It's about time Limetech pulls that release and also gets a new fixed version released to stop this mess. Peter Is there that many? If someone has participated in all and can collect, we can direct limtech to the condensed thread for review and further action.
April 20, 201016 yr OK, I found 3 active and 2 close calls Formatted data drives; http://lime-technology.com/forum/index.php?topic=6084.0 http://lime-technology.com/forum/index.php?topic=6008.0 http://lime-technology.com/forum/index.php?topic=6109.0 Close calls; http://lime-technology.com/forum/index.php?topic=6096.0 http://lime-technology.com/forum/index.php?topic=6016.0
April 21, 201016 yr Here's a fresh one for you: http://lime-technology.com/forum/index.php?topic=6117.0
April 21, 201016 yr This has me concerned. How does a pior data disk get formatted like this. There's gotta be a way to protect people from themselves. I've long felt that the single format button on the first page is a problem waiting to occur. I really think it should be an individual drive option. I completely agree! Or can we at least get the Format and Restore label and instructions to be in BOLD RED? To signify that these options have a permanent repercussion that may result in loss of data... It doesn't take much to make these changes...
April 21, 201016 yr Being one of the members who formatted all their disks... How difficult would it be to create a second page after clicking the format button showing what drives will be formatted and then a go/no go from that point?
April 21, 201016 yr I've reported this to limetech. To be frank, I almost made this mistake a few days ago. What happened is I ran out of memory while doing the par2 testing. I rebooted the machine. It came up. I refreshed the browser for the unraid display with the F5 key. My array came up, but everything said unformatted. I was really looking for the stop button. I did not realize it right away. Right under my mouse was the FORMAT button. I think when the array came up it too too long to do the mounts, so emhttp thought the drives were unformatted. Somehow this has to change. The detection of unformatted vs unmounted really has to be handled better. It's too easy to make a mistake.
April 21, 201016 yr Author A final update....ALL OF MY DATA WAS THERE AFTER ALL....in the lost + found as Joe suggested. Man, was I surprised. Thanks again! As for having additional safeguards to keep me from doing this again, I would definitely vote for that. Not pushing the blame on the software, I chalked it up to a rookie mistake. But we do kind of get used to having to hit an additional button when it asks 'are you sure you want to do this, cause it's gonna have serious repurcussions'
April 21, 201016 yr I think I'm going to write a package with an install script that renames the mkreiserfs. When I "almost" made my mistake, I was looking for the stop button, I almost accidentally clicked on FORMAT It's too easy to make this mistake.
Archived
This topic is now archived and is closed to further replies.