All Shares are using High water but a drive filled while others are half full?

[jango_22]

As I stated in the title all of my shares are using the highwater fill type but when I checked my drive usage one drive is completely full while the others still have like 80% left.

I have a mix of different drive sizes so maybe I just have still not gotten a full grasp on how high water works but I am very confused.

 

My main goal is to have all drives filled to a similar percentage at any time so I don't end up with one drive that has a lot more space used than others. any insight would be appreciated. (screenshot of drive usages attached.)

[BRiT]

Check your split levels and disk includes.

[jango_22]

18 minutes ago, BRiT said:

Check your split levels and disk includes.

All the shares that have more than a gig or two of data were set to "split any directory as required" and to use all disks.

I had a windows backup share which I had not realized was 1.9 TB and might not have been split across many drives so that could have had something to do with it.

[jango_22]

Actually Yeah it appears that the almost 2tb backup (which was incomplete anyway) was only split across my 2 largest drives I'm assuming once the one that was full filled up it started writing to the second. Ill make to add minimum file sizes to my folders now.

[trurl]

8 hours ago, jango_22 said:

Actually Yeah it appears that the almost 2tb backup (which was incomplete anyway) was only split across my 2 largest drives I'm assuming once the one that was full filled up it started writing to the second. 

 

If the backup is a single file then unRAID will not split it at all. Each disk is an independent filesystem so each file must fit completely on a single disk. Possibly it was incomplete because it ran out of space.

 

8 hours ago, jango_22 said:

Ill make to add minimum file sizes to my folders now.

 

This may be what you meant, but I will explain it just in case. What you need to do is set Minimum Free for each of your user shares to be larger than the largest file you expect to write to the share. unRAID cannot know how large a file will become when it starts writing it. If there is less than Minimum Free remaining on a disk, unRAID will choose a different disk to start writing the file.

 

There is also a Minimum Free for cache in Global Share Settings. It comes into play if cache needs to overflow to the array. See here:

 

https://lime-technology.com/forums/topic/46802-faq-for-unraid-v6/?page=2#comment-537383

 

[jango_22]

On 4/23/2018 at 7:43 AM, trurl said:

This may be what you meant, but I will explain it just in case. What you need to do is set Minimum Free for each of your user shares to be larger than the largest file you expect to write to the share. unRAID cannot know how large a file will become when it starts writing it. If there is less than Minimum Free remaining on a disk, unRAID will choose a different disk to start writing the file.

 

There is also a Minimum Free for cache in Global Share Settings. It comes into play if cache needs to overflow to the array. See here:

 

Been a while oops. But yes This is exactly what I meant. What had happened with that windows backup was I had my backups share set to use the Cache drive just like my normal day to day storage shares but the backup I was running was much larger than my cache ssd. I remembered this would cause an issue for the backup about halfway through the ssd filling up so I turned on the mover and then it managed to limp along until I had written close to 2 TB into that folder. (It was an unnecessarily overly comprehensive backup which included my entire downloaded steam library which has no need to backup.) once I realized the faults in that method I set a minimum free space of like 100gb for that share and disabled caching because the cache disk is what had caused those backups to fail in the past not the actual filling of the 2T disk. It has since leveled out more across my drives. (granted I have still eaten up over 4TB and will need to buy more drives soon but alas. 

 

Anyway, Thanks for the help and information!