Best hard disks for Microserver

Best hard disks for Microserver

Author
Discussion

Slushbox

1,484 posts

105 months

Saturday 17th June 2017
quotequote all
Sam. said:
Been running 8 x 4TB WD Blues in my CentOS server for over 4 years now. No failure of drives. Had around 1 hour downtime since it was booted up.

Cant fault them for the price. Im really surprised they havent failed yet
Hard to tell with disk life, we had a five-year old workstation drive eat two drives a month ago, to the point where we couldn't read the SMART data, one was a Seagate cheapie, and the other a WD Blue. Suspect power failures/spikes. Replaced both drives as five years seems 'enough.'

Also have another six year old machine which hasn't been opened since it was built, no idea what's in it. In use daily. We're waiting for the 'Disk Failure' phone call.

Had a look at the SMARTs of the working drive we retired out of the 2010 Dell workstation, Seagate Barracuda with 7910 hours on it. About 1000 8 hour days of 'use'.

Will report back on the replacement RAID1 Barracudas in three years. :-)

Storage warehouse BackBlaze produced their annual Drive Death report in May, it's here:

https://www.backblaze.com/blog/hard-drive-failure-...


Edited by Slushbox on Saturday 17th June 06:38

amacieli

2 posts

82 months

Friday 23rd June 2017
quotequote all
Brother of AngryBiker, here.

The server he has (and I have) is the latest version of Ubuntu server, no GUI, and minimal everything installed. The OS is on one small disk, and we have 5 data drives, 4 TB apiece. AngryBiker, being of the angry sort, has struggled at times with Linux, and I've had to reinstall his OS at least twice. Zero data loss, and re-attaching the data volumes was a trivial one-line command. He stays out of the command line, nowadays. The hardware is a bog standard mini-ITX with an Intel i5 chip of some non-trivial description and 8 gigs of RAM.

The file system on the data drives is just BTRFS - no RAID cards involved. BTRFS does all of the work. I was first intrigued by BTRFS by an article ages ago on Arstechnica, where it described how that file system automatically detected and corrected bitrot, as well as offering some level of protection against loss of a physical drive.

I check our machines at least monthly, make sure all updates and patches are installed. I also run SMART utilities to check physical drive health. If one drive would ever start to die, I'd just have it removed and replaced with a fresh one, and the system would automatically rebuild (if not, again - it's just a one-line command to do the rebalancing). Because drives fail not so often, and it's even more unlikely for multiple drives to fail at once, this is probably a reasonable enough way of reducing data loss. Not bulletproof, but a reasonable effort/cost-to-risk ratio..

As far as my own data is concerned, I also have a spare NAS that I never turn on except to copy stuff to it. Offline onsite backup. All my stuff is categorized as follows:

Personal documents - CrashPlan backs that up. I also have a copy on OneDrive, because I subscribe to Office365.
Media (photos, home video, home audio) - CrashPlan backs that up
Software downloads - CrashPlan backs that up
Movies - my brother has many of the same, so if my machine dies, I can re-rip (or whatever), or get a copy from him. Inconvenient and time-consuming, but not disastrous.
Music - CrashPlan backs that up (all that metadata tagging - not going to re-do that!)

The server also runs CrashPlan directly (it's a bit of a hack, but the CrashPlan support pages tell you how to do it). It also runs Plex, which removes the need for excessive rubbish in the living room etc., because the Plex clients basically run on anything and everything including Windows Phones. It also runs an LDAP server for the "corporate" directory for my VOIP phones.

Works great. I agree with many commenters that these NAS drives seem to last "forever" and not die. However, plan for the worst. Hopefully my strategy above should keep the possibility of data loss at an absolute minimum. I also test that CrashPlan actually works, every so often. Would appreciate thoughts.

essayer

9,067 posts

194 months

Friday 23rd June 2017
quotequote all
I have been really impressed with CrashPlan. We deleted some files at work (4GB worth) then only noticed a year later when the files were needed. Restored them from the control panel without a hitch.

Angrybiker

557 posts

90 months

Friday 23rd June 2017
quotequote all
amacieli said:
Brother of AngryBiker, here.

and I've had to reinstall his OS at least twice.
.
Only because I forgot my password

TonyRPH

12,972 posts

168 months

Friday 23rd June 2017
quotequote all
Angrybiker said:
amacieli said:
Brother of AngryBiker, here.

and I've had to reinstall his OS at least twice.
.
Only because I forgot my password
You don't need to re-install Linux because of a forgotten password.

Simply boot from a rescue CD, mount the root file system of the forgotten password machine, and edit /etc/shadow - just blank out the existing encrypted password and reboot - it will then login without a password.

You could also go the full hog, and mount all the required file systems (/dev, /proc, /sys) then chroot to it and simply run passwd - and it'll change the root password.

No need for a re-install.

amacieli

2 posts

82 months

Friday 23rd June 2017
quotequote all
LOL. Agree - but try doing all that over a muffled phone call only when you can't see the screen! smile And the corresponding huffing about sacrificing a thumb drive etc. We also had problems when I was telling him what to type into the command line to update his system. I do that remotely myself, now that his server pings DynDNS or whatever so I know what his ever-changing WAN IP is.

Edited by amacieli on Friday 23 June 14:25