January 4, 201610 yr I've had to reboot my server recently so the data in the screenshot isn't fully complete, but I've noticed a strange behavior recently where unRaid continues to fill up drives that have the least amount of storage available. Notice in the screenshot the number of writes for some recent files I've saved to the array. There is another drive that has considerably more space, but unRaid continues to choose to put it on a drive that is already fairly well-utilized. Wouldn't it make more sense to start leveraging drives that have more space available?
January 4, 201610 yr I've had to reboot my server recently so the data in the screenshot isn't fully complete, but I've noticed a strange behavior recently where unRaid continues to fill up drives that have the least amount of storage available. Notice in the screenshot the number of writes for some recent files I've saved to the array. There is another drive that has considerably more space, but unRaid continues to choose to put it on a drive that is already fairly well-utilized. Wouldn't it make more sense to start leveraging drives that have more space available? All depends upon what you've set up for your allocation method (most free, highwater). Presumably you've got high water set and since your largest drive is 2tb, unRaid will fill up all drive 1 until it is 2tb / 2 (1tb) / 2 (500GB) / 2 = 250 GB And then it will fill up drives 2+ to 250GB at which point it will begin to fill up drive 1 to 125GB and so on.
January 4, 201610 yr From the manual: http://lime-technology.com/wiki/index.php/UnRAID_Manual_5#Creating_User_Shares Allocation method When a new User share is created, or when any object (file or directory) is created within a User share, the system must determine which data disk the User share or object will be created on. In general, a new User share, or object within a User share, will be created on the data disk with the most free space. However there are a set of share configuration parameters available to fine tune disk allocation. The basic allocation strategy for a share is defined by the Allocation method configuration parameter. You may select one of two Allocation methods for the system to use: Most-Free In this method, the system will simply pick the disk which currently has the most free space. High-Water In this method, the system will pick the disk which currently has the least free space that is still above a certain minimum (called the "high water" mark). Suppose in our example above, we have this situation: disk size free disk1 80GB 75GB disk2 120GB 110GB disk3 80GB 70GB The initial high water mark is set to the 1/2 the size of the largest disk; in this case, it will be set to 60GB. In this state, disk1 has 15GB of free space above the "high water" mark; disk2 has 50GB, and disk3 has 10GB. As new objects are created, the system will choose disk3 until the amount of free space on disk3 falls under 60GB. Subsequently, the system will start allocating from disk1 until it's free space falls under 60GB. Then it will allocate from disk2 until it's free space also falls under 60GB. Once the amount of free space on all disks is below 60GB, a new high water mark is established by dividing the old high water mark by 2. The advantage of High-water method is that when writing a series of files, most of the time only one data disk will need to be spun up.
January 4, 201610 yr Author Yikes. I had no idea. I thought I had chosen what was probably the default for unRaid. I've had unRaid for maybe 6 years now. I just checked and I indeed have highwater for allocation method, but the minimum space is 0 KB. That sort of sounds like a recipe for disaster, but I'm not sure. Is there any kind of consensus on what might be the best option here and is it OK to make the change now? I would like to change to what most people consider to be the best option if possible.
January 4, 201610 yr Yes you can change it. set minimum space larger than the biggest file you might write to it so you leave some space in case you modify (add) to a file that's already on the drive. I use my media server, #1, for recording TV shows almost everynight for a maximum of 96GB so I set it to 120GB min space.
January 4, 201610 yr Author Thanks. Currently, I'm pre-clearing a 4TB drive to be my new parity drive. I then plan on taking the old parity drive (2TB) and make it a data drive. Does it make sense to choose MOST FREE for all my shares at that time. It seems to me that would sort of force unRaid to exclusively use the new data drive for some time since it would definitely have the most free space of all the drives and need to get close to 1.5 TB being used before it might look to leverage space elsewhere since every other drive pretty much only has about 500 (or less) gigs free. My logic for doing this is because my most watched media will be new content. This will force the new data drive to soon be almost the only drive spun up for reading and writing (excluding parity). The rest of the drives would only be used once in a while as I look to access older content (infrequent). Also, while I'm on the topic of how unRaid saves data let me ask one other question I should really know the answer to already, but I don't. I notice that if I bypass the shares and just look at what is on each specific Disk (Disk1, Disk2, etc.) itself there is a folder structure. Is it OK and common practice to save files directly to the Disk? This would essentially allow me to force data on to a specific disk, but the shares would still see it properly, and parity is used to ensure the data is protected....correct?
January 5, 201610 yr Community Expert ... Also, while I'm on the topic of how unRaid saves data let me ask one other question I should really know the answer to already, but I don't. I notice that if I bypass the shares and just look at what is on each specific Disk (Disk1, Disk2, etc.) itself there is a folder structure. Is it OK and common practice to save files directly to the Disk? This would essentially allow me to force data on to a specific disk, but the shares would still see it properly, and parity is used to ensure the data is protected....correct? A user share is just the aggregate of all top level folders with the same name, so yes you can write to them directly. Just be careful not to move or copy from a disk to a user share or from a user share to a disk or you could get the User Share Copy Bug (search if you don't know about it).
January 5, 201610 yr Yes you can do that but don't ever copy files from the share to a disk or from a disk to a share, only disk to disk. Or if 2 different shares, share to share.
March 20, 201610 yr I have a related question... I have a user share that is comprised of two disk shares. Somehow allocation method got set to high water - not sure how, don't really care at this point. But one of the disks has 0kb free... And now that user share, even after I change it to "Most Free", still shows the user share at 0kb free. I even removed the disk that is completely full from that user share, where the other disk share has 8.8gb free... The user share still is showing 0kb free. I'm on 6.1.3. What might I be doing wrong?
March 20, 201610 yr Community Expert I have a related question... I have a user share that is comprised of two disk shares. Somehow allocation method got set to high water - not sure how, don't really care at this point. But one of the disks has 0kb free... And now that user share, even after I change it to "Most Free", still shows the user share at 0kb free. I even removed the disk that is completely full from that user share, where the other disk share has 8.8gb free... The user share still is showing 0kb free. I'm on 6.1.3. What might I be doing wrong? Not entirely sure why share is showing 0 free, but here are some things to consider. High water is the default so if you didn't change it that is what you got. User share settings only apply to future writing, so even if you have removed the disk, if it still has a folder for the share then it is still part of the share for reading. Split level takes precedence over allocation method. Minimum free space should be set larger than the largest file you ever expect to write to the share. 8.8GB free is practically full anyway so I think you need to make room. unRAID will not move anything for you.
March 20, 201610 yr Community Expert unRAID will never automatically move existing files between data disks, so you need to take manual action to free up space on the full disk. You might find the unbalance plugin to be off use for a GUI method of moving files between disks.
Archived
This topic is now archived and is closed to further replies.