Has anyone ever experienced the failure of the CPU fan on a server? Does any know how long roughly until a server shuts itself down after this happens if the CPU is moderately active??
I know this is something that involves a lot of variables, but I am curious about a ball park figure from anyones experience assuming all other cooling is functioning normally and a well cooled data center.
IT is dependent on the CPU. power usage is something that was optimized a lot the last year or two, older CPU's will suck up more power when idle.
Proper servers do NOT have CPU fans for exactly that reason. Pretty much all my servers (in racks) have:
This gives redundancy and the vents are part of the case and hot swappable without shutting down the server. They also have redundancy in - there are multiple because they have to cover a wide area to start with, and the server rack case is not high ;)
P3 class, we had a machine that ran with no fan for almost 1.2 years. It wasn't a mission critical machine and after a few weeks became more of a curiosity. P4 Celeron/3.0 SMT, weeks under light load.
All of our recent machines have dual squirrel cage blowers capable of handling a single fan failure.
All of these were in a data center with adequate cooling, hot/cool rows. Most new linux kernels will throttle the machine if it goes into a thermal alert, look at dmesg output to see if it has been throttled.
Get yourself a "spare" server in your nicely air-conditioned server room, run it up to load, grab some sort of monitoring (lmsensors or an infrared thermometer) and pull the CPU fan... :)
I've had a CPU fan on a server that was broken for several years with no harm.
But that was a p166 10 years ago. This information might be a bit outdated. :-)
The server is still running. I don't know if the replacement CPU fan is working anymore.
