Nedry

FWIW, I bought the 2-port version of that IO CREST card (with the JMB582) earlier this year and it has worked flawlessly. I only needed to add a couple of drives, and it seemed prudent to pick the solution that would only use my x1 slot and leave my remaining x4 and x16 for something more substantial in the future. It also generates less heat and uses less power than an HBA. Whole process was as close to plug-and-play as you could ever hope for.

Are there any potential drawbacks I'm not seeing to using this method to bypass the cache drive for USB transfers? Occasionally I will move large amounts of finished media to the server, and under normal settings this will lead to dumping the data to the mirrored NVMe cache and then later to the HDDs. This seems wasteful, in terms of NVMe i/o, because the cache is serving no real purpose here. I know I can disable caching for the share, and reenable later, but I'm hesitant to be continually messing with the mover settings like that, especially if they may require an array restart to take effect. So... can I just use the WEBUI filesystem to copy the files directly from my attached SSD drive to the appropriate user0 folder and bypass the cache that way whenever it's necessary? Because that sounds like a far better solution than toggling mover settings every time.