Evacuation ability clarification

[Beaker1024]

I've done a search on this forum and read the blogs & manuals that I could find that mention DrivePool's evacuation but I don't see a specific statement to clearify things for me.

 

Can you tell me if the DrivePool Evacation balancer has the ability to move files off a failing (for what ever reason) HDD that are not within the actual "pool" itself? [the "other" data]

 

I have left my WHS2011 client computer backups outside of the Pool as I found too many threads / users of DrivePool have thier client backups need to be wiped out and started over to get it into the pool. 

 

The only feature I really want of drivePool for my client backups are to be moved out (Evacuated) if HDD has issues.  NOTE: I do have the HDD that has the client backups share outside of the pool (as "other" data) added to the DrivePool.

 

I hope that makes sense.

 

Thank you for your assistance!

[Christopher (Drashna)]

Specifically what is being referred to is the "StableBit Scanner" balancer in StableBit DrivePool.

 

This balancer will automatically move pooled data off of any disk that is marked as "Damaged" (unreadable sectors) by StableBit Scanner.

 

Because this is a balancer for DrivePool, this only affects pooled data. 

 

 

As for the threads about issues with the client computer backups on the pool, this usually has to deal with a failing/damaged drive. And unfortunately, this has much more to do with WHS's backup engine. For instance, the 4k "commit.dat" file. If that gets corrupted, the service will fail to start. But delete it, and you're fine. Also, it stores the actual data in 4GB chunks. Because the data is deduplicated, one file missing means that you could lose the ENTIRE database (speciifcally, a critical section, like part of the Windows directory). 

 

Because of this, we HIGHLY recommend enabling duplication on the Client Computer Backup folder. That way, if one disk fails, the database should be intact. 

 

Additionally, if you don't want the contents spread out, you could use the file placement rules to limit the contents to one (or two) disks. This way, it's on the one disk, but if Scanner detects an issue with that disk, it should evacuate the contents to another disk.

 

 

 

Otherwise, if there is an issue with the disk, DrivePool can't do anything with the "other" data.

[Beaker1024]

Christopher - Thanks for the detialed response! 

 

I guess I'm a little hesitant to move the "Client Backup" Share into the drive pool as all the searching I have done has found users that do this with existing backups have to wipe out the backups and start over.  Those who started off from stratch with blank client backups and put the share folder in the Pool were ok.

 

Could all of this be circumvented with just deleting the commit.dat file and letting it regenerate? 

 

It'd be nice to have it all in the pool duplicated but I'm definitely very hesitant as mentioned.  BTW I will acknowledge that the 3-4 threads I found saying this issue where on the older Covecube forum (not this current forum) and from about circa 2012-ish.

 

Is this concern outdated due to newer versions of DrivePool? (I'm inclinded to believe there is never an issue regarding drivepool but an issue of WHS client backups be delicate and the act of moving the folder breaks things!)  I don't seem to find this disdcussion in newer forum threads.

 

Thread mentioning a Sticky thread recommending not to put client backup into DrivePool (2012 covecube):

http://forum.covecube.com/discussion/615/client-computer-backups-outside-the-pool/p1

 

[HardForum] 2014 - No real explaination just that it didn't work for them.

http://hardforum.com/showthread.php?t=1802729

"Of note, at least for me, server backup and client backups would not work on pooled drives. I use a separate drive for each."

 

These can definitely be different situations and/or outdated information given DrivePool versions or other conditions.

[RobbieH]

I can concur with Christopher. I have had WHS 2011 backup data get corrupted both inside and outside the pool. And in my case, every time seemed to have to do with a failing drive. I currently have my backups in a pool, and since getting rid of the failing drives, I haven't had any corruption issues since.

 

Also, I have never had to delete all backup data. I did recently have to delete one computer's backup data and reconnect the laptop to WHS, but that was the first time I've had to do this, and it was because a drive failed. The duplicated data was still on the secondary drive, but it was a flaky drive too.

[Christopher (Drashna)]

Beaker,

 

I can't comment about the client backup database issues that he was mentioning because I am pretty sure that he never contacted us about them. 

However, I do know that Windows Server backup won't backup data on the pool, and that is is very picky about the drive used for the backup destination. In fact, I believe that it won't even list the Pool as a potential destination. 

 

 

 

As for the commit.dat file, this doesn't always work, but in some cases it does. I'm not sure why, as I don't really know how it works. Alex could better answer this, as he's reverse engineered the backup database actually (he has a tool that can dump a backup to VHD).

But if you ever have issues, that's the first thing to try. But I would recommend renaming it, rather than deleting it. At least initially.

 

And Paaland nails it though:

 

 

The backup database tends to grow quite large and unless you have enough space to turn on duplication I would not recommend having the backup in the pool. The reason is that the backup database is very sensitive to errors. By putting something in the pool the data is spread across all of the hard drives in the pool. If one of the drives fails you loose the parts of the  backup database placed on that disk. Unfortunately the backupdatabase does not like to loose files and you can end up with a non-working backup-database.

 

If you're using StableBit DrivePool 2.X, you can use the file placement rules to limit which disks the folder ends up. In case you have questionable disks.

That, and there have been a LOT of bug fixes since 2012. The software has come a long way in regards to stability.