Way back in 2023 / 2024 we got massive improvements to FUSE which haven't propagated to unraid. In the year of our lord 2026, are they even at least on the roadmap? can we get some love for the array users who don't want to go down the path of striping? its been the biggest bottleneck for the past decade+ and we have solutions that massive increase performance now, just waiting for unraid to implement.

LWN.net

FUSE passthrough for file I/O

There are some filesystems that use the Filesystem in Userspace (FUSE) framework but only to pr [...]

https://kernelnewbies.org/Linux_6.9#Faster_FUSE_I.2FO

This was all benchmarked back in 2024 and the results were staggering then, and still are now.

https://www.phoronix.com/news/Linux-6.9-FUSE-Passthrough

Can we get this implemented so we can see a 300%+ increase in read speeds and 1000% increase in write speeds?

this one issue is one of the biggest gripes of performance for most people using unraid. adding this solves it and we gain upwards of 10x performance uplift from it.

This has been mentioned before, but according to LT it would require a big code change; it's not just enabling that option (which already is), due to high-level vs low-level FUSE implementations, so while still a possibility, it's not a quick or easy change.

sure but doesn't even the greatest journey start with a single step? adding zfs also was no small feat but it exists within the unraid ecosystem. if the dev ethos is only to look for small flag modification style changes to see wins then that needs looked at.

mergerfs with only a single dev implemented this and people saw significant performance uplift. you'd imagine a paid product, with an entire dev team would have more collective capacity to integrate it.

i really dont want to spend the time to migrate away but it becomes a real problem for a lot of us who have mixed workloads of mostly random read. unfortunately those guys are most effected and would really benefit from a 35x performance uplift.

+1

Seriously, this would be quite needed IMO. If one has to deal with smaller files currently Unraids "default" structure of a nice and flexible array with SSD cache is pretty much worthless. NVMe drives at the speed lower than HDDs, not really great from my experience.

Would be cool if this could be tackled...someday not too far away. Seems like a major improvement.

one solution can also be to use https://forums.unraid.net/topic/144999-plugin-mergerfs-for-unraid-support-topic/

it seems as though unraid / limetech long abandoned JBOD users, so waiting on them to do something meaningful to support them is a lesson in frustration (as it seems most of these feature requests are, theres 1000's of really good ones spanning decades that are completely ignored)

would love to see this as well, even though it might be a complex implementation i think it would be worth moving towards.

On 2/1/2026 at 11:15 AM, JorgeB said:

This has been mentioned before, but according to LT it would require a big code change; it's not just enabling that option (which already is), due to high-level vs low-level FUSE implementations, so while still a possibility, it's not a quick or easy change.

@JorgeB Nonetheless the numbers posted are impressive. Unraid practically has just this problem with the array performance. Maybe LT could give it another thought or is it a definitely off the table?

AFAIK it's not off the table, but it may take some time.