Hey everyone,
I’m running into a frustrating storage bottleneck on my home server setup and wanted to see if anyone else has optimized their pools for this type of specific application behavior. I’ve got a solid TrueNAS deployment handling our standard household containers, media streaming, and local backups completely fine.
However, things start crawling the second I try to spin up a few instances of an Android/iOS mobile emulation pipeline that’s pointed directly at an SMB/NFS share for its working dataset. To give you some context on the traffic profile, I’m hosting a localized sandbox testing setup where I simulate environments to play game sessions configured with high-frequency mobile script executors, automation layouts, and mod profiles.
Because these client tools continuously read and write tiny, highly dynamic asset files and log variations simultaneously across multiple instances, my TrueNAS pool’s I/O wait spikes into the red. The system memory cache fills up instantly, and it causes cascading network latency loops across the rest of our local apps, occasionally even dropping the storage connection entirely.
It seems like the sheer volume of rapid, parallel random write requests from these automated mobile scripts is completely overwhelming the dataset’s standard record size and sync writes.
Has anyone else ran into severe pool degradation or system-wide latency when hosting high-concurrency mobile tools or dynamic client emulators? Should I look into setting up an explicit SLOG (Log device) for this dataset to handle the synchronous writes better, or would it be smarter to adjust the ZFS record size down to match the smaller file sizes being tossed around by the executors?
Any insights on squeezing better random I/O performance out of TrueNAS for this kind of bursty workload would be an absolute life-saver!