July 25, 20169 yr Hello, I have an unRAID Server Plus array (6.1.9) with a 4TB parity drive and 7 other 4TB drives, making 28TB in total. All the drives are HGST NAS drives. Cache (480GB DCS3510) is currently off for the transfer. What I am trying to do is copy a 12TB raw image of my other server's (ESXI/2012 R2, 12TB SSD array+60TB HDD) array to the unRAID system for backup. The copy starts and continues fine until it fills the first disk it hits, then the system doing the copy (Server 2012 R2) gets "delayed write failed" system events. Both systems have X540 NICs, though unRAID doesn't seem to get much faster than 50 MB/s in transfer, limitations of the HGST drives without cache, I know. I'm guessing this is a limitation of unRAID? Is this indeed the case with the file-system? You cannot store files larger than a single drive? Are there any ways around this or will I have to find some other way to backup a 12TB drive image? Thank you!
July 25, 20169 yr Each individual disk in the parity protected array has its own independent filesystem, so no, you can't copy a file larger than the free space available on any single drive. The cache pool, however, has a single BTRFS filesystem if you have more than one drive assigned, so if you have a large enough cache pool, you could use that.
July 25, 20169 yr Author Yep, that is what I figured. I'll probably just use unRAID for user personal file storage then. I'll have to build another server or add storage to the second ESX system here. Thank you for the clarification!
July 25, 20169 yr I was going to ask this same question but found this thread first. I was super excited to use unraid to store some VM backups and act as a seedbox, until I ran my first backup and it froze when the first disk filled up. My backup is of course larger than any single disk. Isn't there a way to format the disks such that they "stripe" the data?
July 25, 20169 yr Isn't there a way to format the disks such that they "stripe" the data? Yes, assign all but one disk to the cache pool. The cache pool is BTRFS, and stripes the data across all member disks. The default is RAID1, which provides full duplication of all data. If you wish to forgo any data protection, you can manually switch that to RAID0. That's not what unraid was designed for, but if you wish, you can force it into that role.
July 26, 20169 yr Why would you have VM images larger than a single HDD? That would suggest you are storing data on the VM image that could be stored on the array. Even my full-fat testing VM barely broke 500GB.
July 26, 20169 yr An alternative is to simply use an imaging utility that will break up the image into smaller "chunks" -- either automatically or as set by you. Then just set the "minimum free" for your UnRAID share to the size of those chunks; and the backups will work without issue. However, it does seem you may be better served by simply using a RAID controller and creating a "drive" large enough to hold your large image file ... hopefully using a fault-tolerant RAID level so you have some protection for your backup.
July 26, 20169 yr Why would you have VM images larger than a single HDD? That would suggest you are storing data on the VM image that could be stored on the array. Even my full-fat testing VM barely broke 500GB. I have a virtual server with a high speed raid array I use for storing my data. I do a lot of work directly from the network, and unraid isn't remotely fast enough for that. Additionally, I'm trying to use Altaro VM backup to easily mirror the VM with my data offsite. It requires local backup storage first, so I was hoping to use unraid for that. The size of my VM is about 5TB prior to Windows deduplication. I know rsync can be setup with unraid, but it is too inefficient for my taste. TL;DR I'm looking for CBT based offsite replication of a 5TB virtual machine.
July 28, 20169 yr WinRAR has the ability to store a file in chunks of user defined size, with or without compression
Archived
This topic is now archived and is closed to further replies.