SOLVED: What happens if my OS drive fails?

[techlife]

As we all know, death, taxes and hard drive failure are the only sure things in life. So, as I have not found a good way to automate a OS drive backup, I'm curious how difficult it would be to get my server back up and running if this happens.

I have 11 drives in the pool totaling 17.3TB of storage. 14.4TB is used. All of my important shares are set to 'uses pool' and 1 extra copy.

I don't know if it's necessary/helpful/useless but right now, I have a share in the pool named 'Backup'. This share is where all of my various PC's automatically backup to. After giving this some thought, and once I began using MySQL to manage my libraries in XBMC, I figured having a backup of it was probably a good idea so I am using the automysqlbackup script mentioned here: http://wiki.amahi.org/index.php/Database_Backup

Is there anything else I should be doing?

I'm comfortable dropping in a new drive and loading the latest version of Amahi, I'm just concerned about getting my shares back up and accessible as quickly as possible afterwards. All of our TV shows, movies, music is stored on the server. 90% of our entertainment centers around my HTPC so knowing this in advance so I'm prepared is very important to me.

Thanks in advance! And if this topic was previously covered and I missed it, feel free to just drop a link for the slow kid

[MyCloudServer]

Hi,


I'm new to Amahi but not new to Linux as I have been a Linux User since 2009 and I have configured my share of Enterprise Servers

First of all, you would need to make some considerations and realizations on what your data is to you

Things like having your data as Nearline data vs Archival data vs Secondary Data Storage (physical native drives on your workstation)

http://en.wikipedia.org/wiki/Nearline_storage

http://searchdatabackup.techtarget.com/ ... -archiving

http://en.wikipedia.org/wiki/Computer_data_storage

In most cases, your data stored on your Amahi Server is at best qualifies to be Nearline Data storage mainly due in fact of the type of drives you use and how it is deployed to you as a solution as opposed to high performance SAS Drives deployed to you as iSCSI and appears to you as a Phyisical Logical Drive

Now since your data is already considered as Nearline Data, then the next step if for you to have archival data storage to Tape Media or to Bluray Discs on Archival grade optical discs and store them offsite

As to your question of backing up files to another Nearline Storage like another NAS or another Amahi Server, in Linux, you do such things by making CRON jobs on mounted storage devices

CRON Jobs can be done on the native machine but stored somewhere else like a relative's house off State with an Online Amahi Server with broadband access (that would be neat if you can manage such solutions)

Or you can subscribe to paid online cloud storage which is quite costly

Greyhole, the technology implemented on the Amahi is quite interesting and more than adequate and resillient for the home user

If so, then address your physical hardware to make it resillient to failure such as putting it in appropriate 6U or 4U rackmount enclosures properly ventillated with fans and redundant power supplies

The thing that makes servers fail is the abuse it takes on startups as this is taking a terrible toll on the power supplies, also, the electromechaical devices spin up and warm up to operating conditions prior to full use

While drives are warming up, algorythms in its firmware track its settings with compensation for linear and volumetric expanson of the platters.

Also note that all electronic and electromechanical devices are subject to mean time between failure rates in terms of power on hours

There will come a time when your drives will fail and reach its end of useful life which make sense for you to keep archival data in the form of Bluray disc back ups of the most important things to you like pictures of the wife and kids

As for the OS drives, it would have been great if you made your build having a separate OS drive maybe 80GB or 160GB in size Raid 1 and have your pool of drives being the WD Green Drives which is known to conserve energy and spin down when not needed or if supported by your system thru Harddisk S.M.A.R.T technology features thereby prolonging the life of the device

In my case I built my Amahi HDA server on brand new parts using a low power core 2 duo LGA775 and the latest LGA G41 chipset motherboard from gigabyte which natively supports 3TB drives with its UEFI bios

With these choices alone, I am addressing reliability and I also mounted quite a few silent running 80 mm and 120 mm fans

I hope I was able to help you a bit

Marl

As we all know, death, taxes and hard drive failure are the only sure things in life. So, as I have not found a good way to automate a OS drive backup, I'm curious how difficult it would be to get my server back up and running if this happens.

I have 11 drives in the pool totaling 17.3TB of storage. 14.4TB is used. All of my important shares are set to 'uses pool' and 1 extra copy.

I don't know if it's necessary/helpful/useless but right now, I have a share in the pool named 'Backup'. This share is where all of my various PC's automatically backup to. After giving this some thought, and once I began using MySQL to manage my libraries in XBMC, I figured having a backup of it was probably a good idea so I am using the automysqlbackup script mentioned here: http://wiki.amahi.org/index.php/Database_Backup

Is there anything else I should be doing?

I'm comfortable dropping in a new drive and loading the latest version of Amahi, I'm just concerned about getting my shares back up and accessible as quickly as possible afterwards. All of our TV shows, movies, music is stored on the server. 90% of our entertainment centers around my HTPC so knowing this in advance so I'm prepared is very important to me.

Thanks in advance! And if this topic was previously covered and I missed it, feel free to just drop a link for the slow kid

[ciscoh]

MyCloud.....very nice detailed answer.

I will offer a much simpler one.

I have done this a couple times due to my habit of tinkering and then having to rebuild.

Rebuild the OS install and Amahi with Greyhole shares/landing zone. Make sure the shares are the same as before


I simply added the drives back in one at a time.......did a "greyhole --fsck", waited for it to finish, rinse repeat....worked great.

[techlife]

@MyCloud - Thanks for the great write-up! Got me thinking about several things that I need to explore further...

@Ciscoh - That's kinda more along the lines of what I was asking. Fortunately or unfortunately, I basically got to find out this past weekend. I had upgraded to F14/A6 and was having several issues as a result. In an effort to repair those issues (and answer this question), I pulled my OS drive, dropped in a new one and did a fresh install of F14/A6. I followed the instructions here as was back up and going about 30 hours later.

[justright]

Thanks ciscoh i think your answer, answered my question I was about to ask. *What happens to the non-default shares after an OS/amahi re-install. What your saying is as long as you remake your non-default shares *with the same case sensitive name* in amahi after a fresh install they should be re-populated during the --fsck, am I correct?

[cpg]

I use rsnapshot, backing things up daily, weeklym monthly and yearly on to a separate drive.

It's a command line app similar to Apple's Time Machine (in fact, the early implementations of TM used this, IIRC).

I wish we had an app with a web interface to configure it.

[repat]

Here is an explanation from the Greyhole perspective, shouldn't be much different in Amahi.

https://github.com/gboudreau/Greyhole/w ... toragePool

Pat

[bigfoot65]

See reconnect existing Greyhole storage pool guidance in the wiki. It outlines the specific scenario.

The Amahi wiki is our primary source for how to's and tutorials for many different topics.

[justright]

Thanks guys, appreciate the replies. Backing up the greyhole.conf and fstab looks to be the go ill look into that.

[bigfoot65]

Is this one solved? If so, I will close the thread.