4 ms·
Ask HN: I'll be buying a replacement hard disk, what should I check?
Hello HN!
My laptop hard drive (rotational, magnetic hard disk) broke just yesterday and I need a replacement.
Since I'm on a tight budget, I'll be looking for a replacement on local offers (craiglist-like websites) and I was wondering:
Assuming I bring my laptop with me, and a bootable Usb drive with some GNU/Linux environment installed (and possibly, additional tools like smart tools etc) what tests should I run upon a hard drive before buying it?
And what are 'acceptable' ranges if an used drive is not just perfect (this is, you know, a very likely situation) ?
Of course I won't be buying second-hand SSDs.
Thanks in advance!
--
znpy
- smt88 12y agoIf you're on a tight budget, I'd suggest a smaller drive rather than a used drive. With Ubuntu, lots of dev stuff (big IDEs, big RDBMS), and a decent number of videos, I'm using about half of my 120GB drive. If you aren't storing/editing music or videos, you should be fine with ~64GB SSD on *nix or Windows 8.1+, and then you'll get the manufacturer's warranty. If you do need to store those things, you should just bite the bullet and buy a cheap HDD. The major downside is speed. You can then save up some money to buy a SSD in the future. Finally, if you have access to one and are in the US, you should buy all electronics with a credit card that offers defect/damage/theft protection (like a standard American Express).
- proactivesvcs 12y agosmartctl --test=short /dev/sdX smartctl --all /dev/sdX Check "SMART overall-health self-assessment test result: PASSED" just under the drive stats. If it fails, the drive is definitely very faulty. Any non-zero of the following would make me likely to reject. If you're in real need and can stretch your standards a little, five or fewer of any could be acceptable. Multi_Zone_Error_Rate Offline_Uncorrectable Reallocated_Sector_Ct Reported_Uncorrect Runtime_Bad_Block End-to-End_Error Non-zero Current_Pending_Sector may be a bad sign, but may also clear itself in time. Treat as above. Any non-zero, but not very large (i.e. obviously misinterpreted) of Command_Timeout. Power_on_hours more than 10,000 means a consumer disk is getting a bit old. If it runs above 45°C idle (depending on ambient) may also mean it's getting old or is worn. Any breached temperature threshold that isn't again, misinterpreted (i.e. 255°C). Any Old_age or Pre-fail attributes that breach the threshold indicate an old/unwell drive.
- znpy 12y agoThanks! Just the answer i was looking for!!!