USB Died and Unable to Read It

[itimpi]

If you want to see the individual drives available as shares you need to enable this under Settings->Global Share Settings.

 

it is disabled by default as if you copy files between a disk share and a user share not realising they can be two different views of the same files you can get data loss as you try and copy a file on top of itself (known as the ‘User Share Bug’).  

[barrygordon]

I am not sure what you mean by the different views and the user share bug?  Can I just have disk shares as I did before and not use user shares, although the consensus seems to feel user shares are better.   Why?

 

I have generally found that better is what you are used to, but I am willing to learn new tricks

[trurl]

User shares are the only way you can have folders that span disks. It also allows you to have writes cached then moved to the array.

 

User Shares are simply the top level folders across all disks with the same name. So, a file is on a disk, and it is also in a user share. But, if you mix disks and user shares then you can accidentally try to copy a file on top of itself, with the result that its contents are lost.

 

I always recommend User Shares, but if you don't care anything about spanning disks, and you don't care anything about caching, then it is fine to keep doing what you were.

 

User Shares are the default, but if you want to use disk shares instead of user shares then go to Global Share Settings and disable User Shares. You should use either disk shares or user shares, not both, since if you mix them when moving/copying you can lose data as mentioned.

[Frank1940]

1 hour ago, barrygordon said:

I did like the ability to put stuff exactly where I wanted it (by disk). Is that still possible? If so How?  On second thought maybe I should forget about Disk shares and just use User shares.  I will need to research that more. 

 

 

The problem occurs when copy is made from a Disk Share to a User Share (and visa versa).  If the file being copied is on the same disk as is involved in the Disk Share, Unraid will be reading and writing to the same file which is a recipe for disaster. You can read about the whole thing here:  

 

    https://forums.unraid.net/topic/32836-user-share-copy-bug/#comment-316512

 

 

It gets a bit technical and, even if you are aware of the problem, it is so easy to make a mistake.  And BAM, the file is clobbered...

 

22 hours ago, itimpi said:

• Use Tools->New Config to reset the array
• Assign your data drives and now since you know the parity drive serial you can assign that as well.    The data drives can be in any order as with a single parity system the order does not affect parity
• Check the box to say parity is valid and start the array
•  

 Notice the BOLD in the quote above!!!!    This will prevent a error correcting parity check from beginning.  After you have assigned the Parity drive and everything looks good, then start a  non-correcting parity check.  You will know within a few minutes if that was not your old parity drive.  (It will be throwing up errors like crazy.  A few hundred errors are nothing to be overly concerned.   If you are getting thousands,  then you need to stop and investigate.  Remember when you run a non-correcting parity check, you are not writing to the parity disk.  It is only checking the results of the parity calculation against the actual contents of the disk.) 

Edited May 8, 2019 by Frank1940

[barrygordon]

Thanks.  That clarifies it.  I never allowed a file to span disks, files were always contained on a single disk. I think I will stick with what I had and what I am comfortable with.  I never used a cache disk and do not intend to start now.

 

The problem with unRaid is that it is too good. I've had uptimes measured in years with the only reason it was stopped is when I upgraded the software.  When something is that reliable you forget about it and when you have to do something that is really quite simple it appears complex. I built my unRaid tower in 2009. It has two PSUs in case of a PSU failure, each of them being hot-swappable;  5 internal hot-swappable fans, and 12 drive bays.  I thought I had it all covered. Having designed data centers at one time in my career I know that the main failure causes of electronics are power issues and temperature issues.

 

Since my other passion is Home Automation, the house has its own "Server room". the Tower lives in that room. The room is individually air-conditioned, kept at 69 degrees F.  There is a 2KW UPS that handles all the electronics in the server room, and the house has a 22KW standby generator that kicks in, or is supposed to, after 20 seconds of no utility power. I thought I was covered.  Well, Murphy's law operated. The generators battery had shorted out (unbeknownst to me)  and so when the power failed in the middle of the night (for 3 hours) the generator could not start. I live in central FL where power glitches, lightning storms, and power failures are not uncommon. The generator is being retrofitted with a Wi-Fi interface that will log to my iPhone all weekly self-test results and immediately message and email me when a specific failure occurs.  I need to remember to take my iPhone off "do not disturb".

 

One of my major interests has always been film.  My family was in the film industry and some still are.  One of my cousins "owned" a small TV show on HBO called the "Sopranos" and later became the CEO of Paramount.  The Tower has over 700 files made up mostly of movies ripped from Blu-Ray and dvd, but also concerts and certain classic TV shows. I designed my own theater, which my late sister declared was not a "Home Theater" but rather a screening room. If You are interested in that stuff, check my web site (the-gordons.net).

 

Again thanks for all the assistance.  The second best thing about unRaid is its community!!

 

Barry Gordon

[trurl]

34 minutes ago, barrygordon said:

I never allowed a file to span disks, files were always contained on a single disk.

Unraid never allows a file to span disks either. Each disk is a completely independent filesystem that can be read independently on any computer that supports Linux filesystems. Each file is completely on a single disk. 

 

User shares allow FOLDERS to span disks.

 

37 minutes ago, barrygordon said:

the only reason it was stopped is when I upgraded the software.

What version Unraid are you currently running? The rest of this post assumes you are on a reasonably recent version of Unraid. You should be if you aren't.

 

38 minutes ago, barrygordon said:

When something is that reliable you forget about it

If you are going to forget it you absolutely must set up Notifications to alert you immediately by email or other agent when Unraid detects a problem.

 

And, you should always keep a current backup of the flash drive. You can always download a zipped backup of your flash at Main - Boot Device - Flash - Flash Backup.

 

 

[barrygordon]

I believe I am done. The drives are in the same numeric positions they were before all this started.  Everything is powered down. I just need to move the Tower case back into the rack in my server room.  I have not yet decided if I want to use disk shares. Can I use disk shares and not user shares?  I did not see any enable/disable option for either one under shares.  The view option under user shares is nice as it shows you what file is where. I am running the latest stable release 6.6.7.

 

The format I use to hold my ripped (.mkv) files is that each .mkv file is named with the title of the movie, and it sits in a folder (not a share) with the same name.  If I go with User Shares, how do I establish the named folder and ensure that the .mkv file is in that folder when I write it.  What concerns me is if I make the folder using SMB, might it end up on a drive that has not got enough space to hold the .mkv file?  Will there then be two folders with the same name on two different drives, one of which will contain the movie?  The Makemkv program will make the folder and write the file to the drive, I am just wondering if it is smart enough to make sure there is enough space on the drive.

Edited May 8, 2019 by barrygordon

[itimpi]

You enable/disable User Shares and Disk shares under Settings->Global Share Settings

[trurl]

4 hours ago, barrygordon said:

might it end up on a drive that has not got enough space to hold the .mkv file?

You would set Minimum Free for each user share to larger than the largest file you expect to write to the share. If a disk has less than Minimum for the share, Unraid will choose another disk.

 

 

You joined the forum exactly one year before I did. I'm pretty certain that User Shares were already a feature of Unraid when you joined.

 

If you don't want to bother with learning all the details of User Shares, then do Disk Shares instead as you have been doing. Just don't do both.

 

User Shares gives you a lot more functionality than disk shares, but if you don't care about that additional functionality then maybe it doesn't matter to you.

 

User shares allow folders to span disks, and if one disk becomes too full another disk will be used. And you can get faster writes by caching them to a disk (SSD even) that isn't affected by the slower writes to the parity array. Cached writes can be scheduled to move to the protected array when you aren't using the server.

 

Here is a wiki link about User Shares that gives a good overview and illustration:

 

https://wiki.unraid.net/UnRAID_6/Overview#User_Shares

[barrygordon]

I will start with disk shares while I research User shares. I believe it is trivial to move between the two and I will only use one or the other but not both.

 

Having forgotten most of my Linux CL skills I tend not to log onto the tower or operate at the CL level.

 

My smallest drive is 2 TB and before I load a .mkv file I check free space and decide where to put it.  From what I understand User shares will do that automatically.  Thanks for the link to the wiki info. I will study it

 

For those interested here are two pics of my 10-year-old server sitting out of the rack. In the open case picture, you can see the two PSU's in the lower right and the internal hot-swap fans at about the midline.  The front (left) shows an optical drive (DVD), An internal HDD which is not used but has WIN XP on it and can be booted, and a general media interface which is not plugged into the mobo. At the rear of the case are two exhaust fans.  The case and its components are very loud, hence it being in the server room and not the theater.

 

[barrygordon]

The array is doing a parity check, is 20% done with no errors noted.

 

Something strange though, Of the 10 data disks 8 show the file system as reiserfs, and two of the drives as xfs.  This may have always been the case and I never noticed it before. Should I be concerned?

 

Once again thanks for all the help!! 

[Frank1940]

XFS is now the default file system for Unraid.  ReiserFS is now unsupported by the original developer.  (For the past seven or eight years, he has been serving a life sentence for murdering his wife.)  During this time period, disk capacity has increased significantly and ReiserFS seems to have real issues with write performance as these large capacity disks fill up.  In fact, many of of us have converted our ReiserFS formatted disks to XFS as a precaution against problems.  

[trurl]

20 minutes ago, Frank1940 said:

many of of us have converted our ReiserFS formatted disks to XFS as a precaution against problems.  

Note, however, that this "conversion" is actually a reformat of the disk to the XFS filesystem. So to convert a disk to XFS, you must move all the data off it first.

 

I just searched both pages of this thread, and discovered that nobody had ever asked for or received your diagnostics. Might be useful for you to post them so we can take a look and see if you have any other potential problems looming in your "set it and forget it" system.

 

If you want, go to Tools - Diagnostics and attach the complete diagnostics zip file to your next post.

 

[barrygordon]

The array seems to be running nicely.  I have set up notifications. It took me a while to figure out what I had to enter for email but I eventually got it and the email test did send me a message which I received.  I looked at various parameters and corrected them (e.g. time).  I dealt with Lime-Technology and now have a properly registered  Flash drive.  As I play with the system things come back to me. I really like the GUI interface.  As per a request in this thread I have attached the current syslog.  I am assuming I can register several flash drives but each flash drive will have it's own registered key based on the GUID. What is the size recommendation for the flash drive?  My current drive is a 23 GB SanCruzer.  

 

My Configuration is as follows:

     10 Data drives, 1 Parity drive, 21 TB of total storage, 1 TB used, 5 TB free, No Cache drive, User Shares enabled, Disk Shares Disabled, daily notifications, spin down after 30 minutes no activity.

 

I do not use a cache drive as writes to the array are only when I update my movie library or do full backups from other systems

 

My only concern is that I ran a parity check which claimed there were many errors. It took approximately 10 hours and reported approximately 4000 errors.  I do not remember it ever reporting errors before.  I tried to read some data and it all read fine (actually just played some movies). I have stuff in the drives I no longer need so I am thinking about just deleting that stuff and rerunning a parity check.  When I run the parity check should I "Write corrections to the parity drive"?

 

Comments, advice greatly appreciated.

tower-syslog-20190511-1144.zip

[trurl]

14 minutes ago, barrygordon said:

As per a request in this thread I have attached the current syslog.

 

Was really wanting the complete diagnostics zip.

 

On 5/9/2019 at 7:24 PM, trurl said:

go to Tools - Diagnostics and attach the complete diagnostics zip file to your next post.

 

[barrygordon]

Oops my Bad.  Need to read more carefully and not jump to conclusions.

 

Attached is what you asked for.

tower-diagnostics-20190511-1235.zip

[trurl]

Is the Dashboard showing you SMART warnings for any of your disks (it should be).

 

Are you getting Notifications for those disks (you should be).

 

[barrygordon]

Yes I am.  Here is what is in the archieved notifications:

 

Time                   Event                                            Subject                                                            Description                                                        Importance

11-05-2019 13:00  Unraid Disk 5 SMART health [197]  Warning [TOWER] - current pending sector is 2  Hitachi_HDS722020ALA330_JK1170YAHU7T6P (sdl) warning

11-05-2019 13:00  Unraid Disk 5 SMART health [5]     Warning [TOWER] - reallocated sector ct is 25    Hitachi_HDS722020ALA330_JK1170YAHU7T6P (sdl) warning

11-05-2019 13:00  Unraid Disk 4 SMART health [5]     Warning [TOWER] - reallocated sector ct is 1       Hitachi_HDS722020ALA330_JK1101YAHM2DYV (sdm)  warning

11-05-2019 11:42  Unraid Disk 7 disk utilization           Alert [TOWER] - Disk 7 is low on space (98%)     Hitachi_HDS723020BLA642_MN1210F33A1A3D (sdj) alert

11-05-2019 11:42  Unraid Disk 6 disk utilization           Alert [TOWER] - Disk 6 is low on space (100%)   Hitachi_HDS723020BLA642_MN1210F339Y03D (sdk) alert

11-05-2019 11:42  Unraid Disk 5 disk utilization           Alert [TOWER] - Disk 5 is low on space (99%)Hitachi_HDS722020ALA330_JK1170YAHU7T6P (sdl)       alert

11-05-2019 11:42  Unraid Disk 4 disk utilization           Alert [TOWER] - Disk 4 is low on space (99%)Hitachi_HDS722020ALA330_JK1101YAHM2DYV (sdm)      alert

11-05-2019 11:42  Unraid Disk 3 disk utilization           Alert [TOWER] - Disk 3 is low on space (99%)Hitachi_HDS722020ALA330_JK1174YAHMA9XW (sdf)      alert

11-05-2019 11:42  Unraid Disk 2 disk utilization           Alert [TOWER] - Disk 2 is low on space (99%)Hitachi_HDS722020ALA330_JK1130YAH6B9LT (sde)       alert

11-05-2019 11:42  Unraid Disk 1 disk utilization            Alert [TOWER] - Disk 1 is low on space (98%)Hitachi_HDS722020ALA330_JK1174YAHLP94W (sdd)      alert

 

The utilization messages are proper as that is the state of those drives.  I assume I can disregard them.

I am not sure what I need to do about the first three messages regarding disks 4 and 5 

[trurl]

Did you get emails for these? Or just notices in the webUI? You can adjust what notifications you get and how you get them. I prefer to get emails for some things so I will know about them immediately without having to look at the webUI.

 

You can also adjust the utilization warning on individual disks by clicking on the disk to get to its settings.

 

I am not usually that concerned about a few reallocated sectors, but the pending sectors on disk5 might be a problem. A pending sector is one that it will try to reallocate when it is written again, but until that happens, it might not give reliable results when read. And that could compromise rebuilding a different disk, since all bits of all disks must be reliably read to reliably rebuild a disk.

 

And, disk5 is nearly 9 years old, as is disk4. I haven't checked the age of the others. Did you have any plans to replace any of your disks?

 

Most of your disks are very full, and they are ReiserFS which may not perform well when so full. It might be time to think about replacing these old disks with new, larger disks. Since you already have a 3TB parity, you could replace these problem disks with 3TB and get more capacity. Or you could get larger parity and have even more room for upgrades of the other disks.

 

These days much, much larger disks are no more expensive than those disks were when you got them. I often recommend people use fewer, larger disks instead of more, smaller disks because fewer disks are fewer opportunities for problems, larger disks are more cost effective, and larger disks are often faster due to increased data density.

[barrygordon]

I did get emails for all the notifications.

I have been thinking along the lines of replacing the older disks (starting with #5) with new 3TB disks. Is there someplace that describes the process to do that.  My system has one spare unused drive bay if that helps.  

 

Plan A - Replace the older disks one at a time starting with #5 with 3TB drives

Plan B - Replace the Parity Drive with a 4TB (or ?? what would you recommend) Then replace the data drives with N TB drives (N=??)

 

I am getting on in years so I am not sure how much longer I will be needing the array.  Its basic use is for my movie collection. Most data is written once (when the movie is ripped) and then read several times when a movie is watched. I do have the original DVD/BR disks for all my movies.

 

You are being very helpful and I appreciate it

 

[trurl]

Maybe replace disk5 with a new 3TB just to get that one out of the array, then get a larger parity. You really need more capacity to begin conversion to XFS, and you can either do that with larger disks, or with additional disks, and I prefer larger disks for the reasons already mentioned.

 

The spare bay is good, but I wouldn't use it up permanently with another array disk. It can be a good place to preclear new disks, or mount other disk outside the array for transferring data.

 

Replacing a disk is very simple. It is really the whole point of parity. Parity allows you to replace a failed disk, but it will also let you replace a disk with a newer larger disk in just the same way.

 

You just assign the new disk to the slot you want to replace and start the array to begin rebuilding.

 

 

 

 

[Frank1940]

35 minutes ago, barrygordon said:

Is there someplace that describes the process to do that.  My system has one spare unused drive bay if that helps.  

See Here:

 

    https://wiki.unraid.net/UnRAID_6/Storage_Management

 

Pull Disk 5 and set it aside.  That is your safety net.  If the rebuilt should fail for any reason (Per Mr. Murphy), you should be able to recover a lot (if not all) of  the files from that disk.  

 

What size for a new parity?  That is a good question.  I personally would recommend (at least) doubling the size of the of the current parity disk.  That way you would gain a lot of additional storage pace as old disks are replaced.  (You have appear to have a lot of very old disks in your array and it is difficult to predict when, which one (or if) will fail.)  By doubling the size of your replacement data disks will get rid of one of the major problems with ReiserFS--- very poor performance of file overhead operations during writes to almost full disks.  (The writes will be directed to now virtually empty larger disks.) 

 

I will point out one thing.  Larger capacity disks have about the same annual failure rates as smaller capacity disks.  So an array with the same capacity with fewer larger disks in it has a higher theoretical reliability than one made up of smaller disks.  

Edited May 11, 2019 by Frank1940

[barrygordon]

Things are getting clearer.  My only concern at this time is that the last parity check (writing corrections to the parity drive) returned over 4000 errors. I assume that a new parity check should return 0 errors unless something strange is really going on. I am going to start one later this evening (writing corrections to the parity drive). Hopefully, it will come back with 0 errors.  I will then order two new drives a 6TB for parity and a 3 TB for data.  I then plan to replace drive 5 (2TB) with the 3 TB drive and when that all settles down, replace the 3TB parity drive with the TB drive.  That will give me a 3 TB unused drive and a 2 TB unused drive.

 

Question - what happened to the concept of pre-clearing a disk?

 

This community is FANTASTIC.  Reminds me of the old Pronto PRO Professional community on Remote Central where I was very active (my younger days)

[trurl]

You must have valid parity before attempting to rebuild any disk. And parity is not valid unless sync errors are exactly zero.

 

About preclear

 

Unraid requires a clear disk only when you are adding a disk to a new slot in an array that already has valid parity. This is so parity will remain valid, since a clear disk is all zeros, and those zeros have no effect on parity. Note that a clear disk isn't needed when replacing a disk, since the disk will be completely overwritten from the parity calculation regardless of what is already on the disk.

 

On old versions of Unraid, if you added a disk to a new slot in an array that already had valid parity, and that disk was not already clear, then Unraid would take the array offline while it cleared the disk.

 

Preclear was invented to clear a disk before (pre) adding it to the array, and so avoid the clearing time with the array offline. Some additional testing capability was added to preclear, so people often will preclear a disk even if they don't need a clear disk, just to test the new disk before trusting it in the array. Some even do multiple preclears as a way of "burning in" the disk to get it past "infant mortality".

 

Newer versions of Unraid will clear a disk when a clear disk is required, without taking the array offline. So the original purpose of preclear is no longer needed. But people will still use preclear for testing and burning in a disk.

 

It is possible to test new disks in other ways, and even in other computers, such as using the disk manufacturer's diagnostics utility often provided as a free download.

 

[barrygordon]

First of all, let me reiterate my thanks.  I am not a beginner in this field (computer science) having been in it since 1960. I have been a programmer, software designer, hardware designer and almost all other subfields of CS.  I was an adjunct lecturer in Computer science graduate school at a major university. I was part of the original GPS team in the mid-1960s, I retired in 2005 as Director of IT operations for a fortune 100 company. I wrote my own HA systems in node.js and built all of the control systems for my Home Theater.  I have always treated unRaid as an appliance. I am not fluent in Unix/Linux but can get by. I believe I do understand how things work at the lowest levels.

 

I have run a parity check 3 times over the past week, always with write "corrections to parity" checked. It always comes back with parity valid and finding between 4000 and 4200 errors. I understand why it believes parity is valid if the corrections were written to the parity drive.  If parity is valid then why will a parity check run again a short time later report the same scale of error (4000+). It is as if parity was not being corrected on the parity drive or some other disk(s) is failing constantly. I want to start replacing the older 2 TB drives with larger 3 or 4 TB drives and the parity drive with a 6TB drive.  I understand why one should not replace a failed drive if parity is not valid. My plan was to replace disk 5 first as that seems to be reporting the most warning notifications, then replace the parity drive with a 6 TB unit, and finally replace the older 2TB drives with newer 3 or 4 TB drives.  I understand the advice "do not replace a failed drive unless parity is valid".  I do know how the parity system works.

 

Because of the errors being reported on a parity check I am not sure what the best approach would be. What information can I provide for someone to assist me.  I don't want to waste everybody's time, but I do want to get this resolved. 

 

The instructions for replacing a failed drive appear straightforward, e.g. Replace a Single Disk with a Bigger One

    This is the case where you are replacing a single small disk with a bigger one:

    Stop the array.

    Power down the unit.

    Replace smaller disk with new bigger disk.

    Power up the unit.

    Start the array.

 

I am surprised that there is no need to assign the new drive into the array

 

 

Thanks.