How to tell if my dedicated server HDD is failing ?
 

read the title .

i have 3 dedictaed servers , non managed.

i was wondering, how to tell if hdd is starting to fail ?

what are the signs ?


thanks for your time.

not going to pretent that i'm S.M.A.R.T., so you can try this: https://superuser.com/questions/1711...f-a-hard-drive
:2 cents:

There will often be no signs. If uptime is important, you need to do both of these:

1. Redundancy (eg RAID) : to allow your server to continue operating when a drive fails.

2. Backup: to allow recovery if the RAID array fails, or someone maliciously deletes your Apache install.

There is little point trying to predict the failure of a drive.

thanks, i have one dedicated server with raid soft 10
or something like that, 2x4tb

i was under impression, if one drive fails then i have to ask the support to switch to another working drive from RAID, but if its automatic then no worries ., i guess.

If the server is unmanaged they may be relying on you to monitor the drives, and request any necessary rectification yourself. Don't assume they'll automatically know when something fails - ask them for clarification...

Don't forget backups.

True, often when it start to show signs data loss already start to occur.

There may not be signs but if there are, sometimes the server will start acting up - mysql / databases not connecting is usually the first sign. Maybe a sudden need for reboots.

Redundancy via RAID is constant, there is a RAID monitoring daemon. If you had a managed service your host would be doing that, if not you're responsible for doing that yourself. Once a drive fails, such monitoring would show you a drive has burned out, but the server would continue functioning using remaining drives. RAID makes things easy as drives are typically hot-swappable - rip out the bad, put in new, rebuild the array.

This doesn't mean that RAID is the same as a backup. If you accidentally delete something on your server, the backup is where you'd be able to get that information, it will be gone from the server. There are also other things that can fail and in this case, RAID wouldn't help, so regular backups are a must.

Half the time a drive crashes it's not even the drive. Just this morning I woke up to no operating system on my computer, swapped SATA cable and it was fine.

I binned (well, put aside) two SSDs because I thought they had reached the end of their useful life. They tested fine in another machine, but when they were put under load in the main server they dropped out after a few seconds of ZFS pool rebuilding. They have a very high TBW value, well beyond the warranty, so I figured the cells were done and some couldn't be written to any more.

It turns out that there was something odd happening with the SATA port or hotswap bay... after a reboot they worked fine, and are still working fine now.

I discovered this AFTER I'd already purchased replacement drives... :mad:

Great point!

If you knew all of the crazy shit that goes on inside of a mechanical hard drive, you'd be amazed that it didn't fly into a million pieces after 10 seconds of operation.

I've been through a couple of hard drive failures and there was no audible warning from the hardware itself. But, I did start to notice weird shit with the daily operation in the OS. Things like files not copying properly or giving errors...or file corruption happening with increased frequency. When you see that, it may be ready to shit the bed.

PLease back up yo shit!

Great thread! Lots of good technical advice and experience sharing in here. Always RAID. Always keep a backup somewhere else. SMART monitoring. Sometimes there is no warning. Sometimes a worn/old drive with few errors simply fails under load. Statistically for most enterprise drives, you can expect failure rates of less than a percent to a few percent. Consumer drives are often much worst. Stuff just fails, so plan for failures and be diligent about your preparation.

...with a managed host like us... (cough)... just don’t worry about it!

Brad