My StartOS node is consistently showing increasing disk writes over time. It is running in a VM on a TrueNAS Scale host (Debian-based with ZFS as the underlying file system) with 48 GB physical RAM. The StartOS VM is the only thing using the assigned SSD. It has done IBD once, and otherwise is running only Knots/DATUM/electrs/mempool. The SSD is already showing 28% endurance used after only ~66 days uptime. I thought it might be insufficient RAM so have tried increasing from 8 to 16 and now 22GB of RAM assigned to the VM. StartOS typically reports only 20-25% RAM used (of 22 GB) at any time. The RAM increases have noticeably reduced the SSD read activity, but the disk writes consistently show this increasing trend over time irrespective of the RAM size. Any ideas where I could adjust settings to reduce this would be appreciated, otherwise it seems this SSD is not long for the world.
16GB is fine for most uses. I don’t know anything about TruNAS, how there metrics are calculated, or how accurate they are. There’s aren’t any specific write optimizations available or supported withing StartOS. I can tell you many people run a StartOS setup, on bear metal, similar to yours, and the write endurance does seem to be a problem. I’m not sure if there is something specific to the VM setup, and the relation to the host that’s causing excessive writes, or if the metrics are wrong. Hopefully someone more familiar with TruNAS may have more insight.