Closed matthuszagh closed 6 months ago
There is some more discussion at https://github.com/kdave/btrfs-progs/issues/610, and a patch for preferred-(meta)data: https://github.com/kakra/linux/pull/31
Thanks! I'll close this, as it's a duplicate issue then.
This is a question. I apologize if it's been asked before, or if this is the wrong place to ask it, but I wasn't able to find an answer elsewhere. If that's the case, please feel free to direct me elsewhere.
In a multi-device filesystem (data single, metadata dup), how does BTRFS handle on which device to allocate a new data block? I seem to recall reading somewhere that it was the device with the largest free space left. Is that correct? Is there a way to give BTRFS a priority of devices, such that it fills up the highest priority device before the next, etc?
I have a desktop computer with 2 very fast NVMe SSD drives and several very large HDDs. Unfortunately, the BTRFS is filling up the HDDs and ignoring the fast SSDs, greatly affecting the performance. Is there a way to resolve this without removing the HDDs from the filesystem?