I can’t find any information or consensus on how bad this kind of temperature is for hard disks. Was setting up a backup server and set the supermicro bmc fan control to optimal, which decided to turn off the fans behind the hard drive since the rest of the system was doing fine (wtf? Is this normal?). It was probably like this for 3-4 hours since I was remoting into it rather than doing things on the monitor. I checked and it gave some temp warnings, so I used smartctl -a /dev/sdwhatever and saw that a disk was at around 68c and shut it off immediately. Cooled down the disks and set fan control to full speed and now they are at a relatively normal 43c.
Consumers drives are probably not too happy about that temp but enterprise should at least survive for a short period of time how ever it will shorten lifespan quite a bit at those temps and it’s usually out of spec (most enterprise drives can do ~55-60C tops ambient during operation).
Yeah I figured. No real data anywhere on this that would give me estimates for how much it would shorten the lifespan unfortunately. I wonder if there is a study on this out there anywhere, because I would love to see it.
If it was running like that for weeks I’d be agitated at finding out, but a few hours? No big deal. If it was so fragile (platters warping or demagnetizing or something), you’d already be seeing problems.
It’s actually colder temperatures that are/were bad for HDDs. A few things to note:
Drives at that time typically had a lower rated operating temp than today, something like 50-55C operating. Most modern drives are rated at 60C. I think I recall seeing some with a 65C rating before.
The drives in question were operating at roughly that temperature constantly
While the graph cuts off at 52C, even if it skyrocketed back up, I’d be willing to bet where you end up is no worse than running cold (though I suspect modern HDD drives are better at that as well, having an operating temperature down to 5C now).
Thanks for alleviating my fears back then. Great source
It has been a bit, and so far I have experienced no failures. They all appear to be spinning like tops. It might change in the future or have significantly reduced lifespan, but for now they appear to be all good. I don’t particularly care about what is on them anymore either since I have moved to 5x 7.68tb sas ssds in a raid with backups for anything I actually care about, as well as 3.2tb nvme ssds for boot. I finally made the swap to nvme for the boot disks because I managed to kill 2 1tb sata ssds with my server within only about 5 months lmao. Fun fact, the proxmox docs are… not wrong, just not right, about how to replace a boot ssd with zfs. Lol.
I had all my fans stop in a fairly densely packed chassis full of discs. The discs were up to 100 degrees when I noticed, which was due to errors. Some of them failed but I had some other issues around that time so it’s hard to say what the cause was or if they actually failed, but most of them are still running today (this was probably four years ago).
Prolly just fine, check the datasheets. As @Log mentioned, current gen operating temperature ratings are commonly 60 C (Exos, Ultrastar, N300) or 65 C (IronWolf, Red). I know from a similar experience with at least IronWolf an overtemperature SMART error gets logged, though, and it’s possible that could be used as a reason to deny warranty.
This failure path is why I like to put fans with fairly high minimum RPM on 3.5s, so even if control cuts out and sets 0% PWM there’s still airflow. Doesn’t work if the control fail flips to DC and turns off the power but that’s a fairly extreme fault. Couple decent choices are PCCooler F5 R120 (500 RPM) and XPG Vento Pro 120 PWM (900 RPM).
I don’t know of any good formal aging studies but, from the somewhat random collection of results I’m aware of, if you wanted to rate 3-4 h of 5-10 C over temperature as 6-8 h of wear life that’s probably on the conservative side.
