hdparm on NVMe disk
To test Trim I am using a method where I create a test file, read the hdparm of the file (with adding the right offset from my dmsetup table and fdisk), but when I try running hdparm on a NVMe disk I got this issue from hdparm, curious to know if it is working for anyone?
hdparm --read-sector 1330176 /dev/nvme0n1
reading sector 1330176: FAILED: Inappropriate ioctl for device
Spindown: is passing -B 1-127 really necessary?
I have tested this with my two Samsung drives, and they both report APM to be disabled. However, setting the spindown time to any value works. Are there drives that really only spin down if APM is enabled/set to 1-127? Soukyuu (talk) 16:53, 22 November 2016 (UTC)
- That's exactly what Lahwaacz (talk) 17:10, 22 November 2016 (UTC) says. Why do you conclude that Samsung is not the exception? --
- Having to sacrifice performance (that's how the manual makes it sound) to enable spindown doesn't make much sense to me. It might as well have been that whoever wrote it had HDDs that were an exception and required setting the APM for the spindown to activate, though. I can't check other drives at the moment, but I don't remember seeing APM being activated/set on windows, yet HDDs still spun down. Soukyuu (talk) 17:22, 22 November 2016 (UTC)
- I have a Seagate desktop disk (ST4000DM000) which also happily seems to obey the spindown timeout even at -B 254. In fact, at -B 127 it loads/unloads the heads practically nonstop, which isn't really usable (nor a good idea). I suspect when the manpage talks about enabling spindown it's only referring to what happens before the timeout interval is reached. In other words, at -B 127 and lower, the drive can make decisions whether to spin down independently of the -S timeout value. (If this is true, the manpage ought to be reworded because this is causing a lot of confusion and misunderstanding.) Either that, or apparently many drives just don't actually follow the specified behavior. TravisE (talk) 19:40, 11 March 2019 (UTC)
- I considered that, but thinking about it, I don't actually know if the page is technically incorrect. I've seen other inconsistencies between drive behavior and the manpage but have no idea if this is a case of a manpage error, other software interfering, or if some modern drives just don't follow specs to the letter. TravisE (talk) 17:05, 1 May 2019 (UTC)
missing command parameter (hdparm after suspend)
is the apm.service correct?
"Putting a drive to sleep directly after boot"
ExecStart=/usr/bin/hdparm -q -S 120 -y /dev/sdb
"APM level reset after suspend"
ExecStart=/usr/bin/hdparm -B 254 /dev/sda
254 is the least agressive value and when i tried it this way the HDD didnt go to standby. when i tried the code from the boot section it worked perfectly.
- You can't expect that all parameters always work the same way - while something works on your hardware, it may not work on other. So there is no universally "correct" solution.
- Likewise, the wiki does not tell you that you must use exactly this unit - users are expected to evaluate if they need it at all or adjust the parameters to their system.
- -- Lahwaacz (talk) 14:46, 21 December 2016 (UTC)
Possible article restructuring (and renaming)
Now almost the whole article is devoted to the power management. It even contains sections for specific vendors (such as WD) and contains too few info about other hdparm features.
I suggest to rename this article to "HDD Power Management" (or so) and include extra information. For example, the whole Seagate brand is not covered however there isAUR toolchain for management of new-generation power saving options for these drives.