I have a server motherboard in a large tower case with several hard disks installed. It's basically a media server that serves over CIFS.
I'm using the 'sensors' command on Linux to monitor the system temperatures. It typically runs around 70c, but if it gets a tough load (downloading several things, hashing the downloaded content) it can cause the fans to spin up to max. An first this would happen briefly, then the fans would spin down after the system cooled. Now, however, the fans never spin down and just howl at full speed forever until the system is shut down for a moment then booted again. Rebooting vs hard cycling the system, curiously, often results in the fans never leaving the "bad state" and they stay on full power.
Troubleshooting steps I've tried:
- Alter BIOS temp control settings
- clean and reapply thermal paste
- double-check fan header connections
- clean dust off internal components
I'm using:
- CentOs 6.8
- Tyan S8230 motherboard
- (2) 12-core Opteron CPUs
- (2) Small Noctua case fans
- (2) Generic, medium case fans that came with the case.
- (2) std CPU coolers that came with the CPUs
ipmiutil sel list output: http://sprunge.us/bJg0gM
Here is how I interpret from
ipmitool sel list
:I think that you have a broken fan that runs slow and that makes other fans to spin up to compensate. Check if you can see all the fans spinning at same speed.