Questions: SMART, predictability, concurrent scans, recovering files and other

[cryodream]

I have been trying out Scanner for the past 3 days. Love it. Tried it about a year ago, it couldn't get SMART for all my drives from the HBA and the external enclosures. Now it does. Woot woot.
I have a couple of questions though:

1. How safe is having no duplication in the DrivePool while running Scanner? I know there cannot be a definitive answer, but just from experience... How much chance is there that scanner will be able to warn me before the drive goes R.I.P.? I am asking for the very cost conscientious reason: do I or don't I duplicate my 50TB+ movies collection? Movies are pretty easy to replace. Although, it takes time and some tedious work.
Case A: If there is little to no chance that scanner will warn me in time, I may need to bite the bullet and buy a bunch more hdds, which gets too expensive quite fast... At least I save on time and avoid potential headaches.
Case B: Scanner is quite good at predicting disasters and I save a ton of money. That would be awesome.

2. Why is there no indication of SMART warning in the list of drives? Is the system tray notification the only place for it, and if so, is it reliable 100%? I'd hate to miss some disaster warnings.

 

3. Most of my drives are in 4 external 16 bay enclosures, connected via 4 external mini SAS SFF-8088 cables to the LSI 9201-16e HBA, which to my knowledge has 4 dedicated 6Gb/s links, one for each port. In this case scanner should be able to run 4 concurrent scans on the HBA without penalty, right? Please correct me, if I'm wrong.

 

4. Which option in the settings is for setting more than one concurrent scan job per controller? Advanced Settings > Configuration Properties > ScanMaximumConcurent ?

 

5. Is there any way to set "drive groups", so that scanner would not run concurrent jobs on the same link?

 

6. Or maybe scanner is smart enough and is able to test and auto-magically detect which drives to scan when, so they would not bog down one another? I see something probably related to this in Settings > Throttling > Do not interfere with other disks on the same controller (also Clear performance data..., whatever that means).

 

7. How does "ignoring SMART warnings" work? I see there is an option on the system tray popup window: Never warn me about SMART warnings/failures. I guess this one is pretty self-explanatory - I check this and never ever get warned about SMART issues, yes?

 

8. There is a button Ignore warning... at the bottom of the SMART popup window. What does this one do exactly? If I press it, would I ignore only current warning(s) or all future warnings for this drive? What about this same warning, but getting worse, eg: increased pending, reallocated or uncorrectable sector counts?

 

9 . About bad sectors. I have one drive's SMART reporting:

Reallocated Sector Count: 0

Current Pending Sector Count: 1

Uncorrectable Sector Count: 1

What should I do? I have read many people opinions, that when u start getting bad sectors, it's a slippery slope, kinda like snowball effect and the drive is pretty much considered doomed... How true is that? Should I bin it? If no, how do I fix it? Full format or chkdsk /r or maybe scanner is actually able to fix it? I moved all files but 1 from the drive already. 1 was giving me cyclic redundancy errors. I was thinking of using SpinRite to try and recover it, if you have any suggestion for a better way to recover the file, please let me know.

 

10. Scanner warned me about another 2 driver with way more severe sector counts:

Reallocated Sector Count: 0

Current Pending Sector Count: 331

Uncorrectable Sector Count: 0

and

Reallocated Sector Count: 0

Current Pending Sector Count: 1464

Uncorrectable Sector Count: 1025

What is confusing, is both drives, even the second one shows the same messages for all sector count parameters: "The drive is operating within manufacturer specified tolerances." What the hell? Does this mean these drives are considered in good condition and usable? I was not able to move ~30 files from the second one, cyclic errors. Gonna need to try and recover also.

 

11. From reading about scanner in this blogpost: Why using StableBit Scanner is a good idea, I gather scanner has some ability to try and recover unreadable files. How does that work? If it succeeds, where does it put the files? How do I do it? Is there a special command/task to launch? Should I try spinrite if scanner fails to recover? Or is it futile?

 

 

 

Regarding questions 5 & 6: if it is not possible to set drive groups and scanner is not smart enough on it's own, I would very much like to add a request for this feature to be added to scanner in the future. I am thinking in the future terms, having 60+ 6TB or 8TB drives on one controller, scanner won't be able to complete scans on all drives in a period of 30 days...

[Christopher (Drashna)]

Lots of questions here!  I'll try to answer everything! If I miss anything, let me know.

 

1. Well, it's always best to run duplication. That way, if a disk fails, you can just remove it from the pool and DrivePool will reduplicate.
   Otherwise, if you're running both, as soon as StableBit Scanner detects unreadable sectors (during the monthly surface scan), it will cause StableBit DrivePool to automatically start evacuating the contents of the disk.  This may not save all of your data, but it may help prevent losing any more. 
   You can optionally configure this to happen for SMART warnings as well (but this is off by default, as temperature warnings will trigger this behavior, as well as some potentially non-critical warnings, such as high  Load Cycle Count warnings).
   
   As for duplicating or not ..... I highly recommend it. Yes, it costs more disk space, but a while ago, I lost ~12 disks (external drives, shelled, and outside of their warranty period).... Not all at once, but in rapid succession.  I did lose some of the data, and I still haven't rebuilt all of it.  And I've had to re-rip and re-download a lot of content that wasn't duplicated. And the older stuff is, the harder it is to find....  
   So if you're okay with that, don't duplicate.  Otherwise, definitely do.
   Additionally, there is an advantage to duplication: Read Striping. 
   
    
2. In the UI, click on the "Settings" button, and select "Notification Settings". There are email, SMS, voice, and mobile notification options here. So, there are a list of ways you can get notified of an issue.
   
    
3. Yes.  More than that.  IIRC, each cable actually has 4x 6gpbs lanes. So each lane could handle 4 drives scanning at one time... depending on how everything is setup.
   
   So you may want to check out the advanced configuration.  You can enable multiple concurrent scans:
   http://wiki.covecube.com/StableBit_Scanner_Advanced_Settings
   Find the "Scanner" section, enable the "ScanWithDirectIO" option, and set the "ScanMaximumConcurrent" to 4 or 8 or whatever you want (or 0 for unlimited). 
   Also, in the Scanner Settings, make sure that you disable the "don't interfere with scans on the same controller" option. 
   I use 8, and it scans 8 of my drives at the same time (a PCI-e 2.0 8x card can handle ~35 drives scanning at 100MB/s at the same time).
    
4. See above. Make sure you scan with DirectIO as well, and disable the controller throttling option.
   
    
5. That throttling option does prevent it from scanning drives on the same controller. So yes, that's what you'd want here.  However, this doesn't work with the concurrent setting (rather, it over rides that behavior.
    
6. See above. And yes, based on activity of the disk or controller, StableBit Scanner will automatically throttle a scan to prevent it from interfering with the normal disk performance.  And over time, the scans may become very staggered (as the 30 day period is determined "per sector", not disk. 
    
7. If you open up the UI, click on the disk in question, you can show the SMART data for the specific disk. (It will also link you to disks with errors). From there, the SMART Details window lets you ignore just the current values (so it will notify you again if it increases) or to permanently ignore the current values that are generating errors.  Also, if there are ignored errors, you can "un-ignore" them there, as well. 
   
   Though the option only appears if you're seeing errors.
    
8. See above.
   
    
9. The current pending/uncorrectable sector count means that it's having issues with "unstable" sectors. Writing to the sectors may fix the issue. There isn't an easy way to do this (as fixing it is "destructive". And StableBit Scanner will likely pick up this sector as an "unreadable sector" during the next surface scan. 
   
   As for fixing it, SpinRite will absolutely do this. A full format or diskpart's "Clean all" command will do this as well (be very careful with this command).  Both write zeros to the entire volume/disk, and may recover the issue. SpinRite basically does the same (writes to the unstable). 
   Though, a "chkdsk /r" may fix this as well. 
    
10. It means just what it says. Each SMART value has a threshold set by the manufacturer. If it's exceeded, then the drive is out of specs and should definitely be replaced. 
    
    However, if you're seeing this, if the drives won't repair (or if the values are rapidly growing), then get rid of them, and replace them.
     
11. If StableBit Scanner detects unreadable sectors, it will prompt you to run a file scan. This will try to determine of any files were actually affected. If so, it uses various head positioning profiles to attempt to read the damaged data. This may or may not fail. 
    
    As part of the file recovery wizard, it does prompt you to specify a location to save the file(s) at. And you can test this out by enabling the "simulate" options.
    
    However, StableBit Scanner doesn't do anything to attempt to recover the sector. The only way to do so really, is to write to it. Since recovery is our top priority, that's what we do.
    But we are/will investigate adding a repair option for the sector (much like what SpinRite does).