DSM-G600, DNS-3xx and NSA-220 Hack Forum

Unfortunately no one can be told what fun_plug is - you have to see it for yourself.

You are not logged in.

Announcement

#1 2010-03-30 22:45:43

switch10
Member
Registered: 2010-02-25
Posts: 27

is my 323 killing hard drives?

I have never had a hard drive fail, in any device, up until about 3 weeks ago.  I had a western digital 1.5 terabyte drive inside my 323 fail.  It was set up as a raid1 with a seagate drive of the same size.  I've gotten the WD drive replaced, and as I was rebuilding the RAID, the seagate drive failed on me.  Now I have to return it as well. 

I guess my question is, has anyone else experienced problems like this?

I think from now on I will be setting up a cron job to rsync the 2 drives instead of a RAID1.

Offline

 

#2 2010-03-30 23:25:19

bjby
Member
Registered: 2009-02-22
Posts: 265

Re: is my 323 killing hard drives?

Generally reading this forum raid1 seems to bring more pain then joy.
If you have the skill to setup rsync mirrored disks then do it.
1. You actually have some sort of backup, oops deleted files dont get really deleted til the sync job run.
2. No hassle with strange raid1 behaivours people tent to run into


The mirror backup has saved me from inconvinience at least once, when fingers have been slippery.

Last edited by bjby (2010-03-30 23:30:07)

Offline

 

#3 2010-03-30 23:58:09

index monkey
Member
From: UK
Registered: 2007-06-14
Posts: 112

Re: is my 323 killing hard drives?

for rsync backup have a peek at the 'rsync time machine backup'.

Last edited by index monkey (2010-03-30 23:59:05)


DNS-323, HW B1, 2 x 2TB WD green, fw 1.08, fun_plug 0.5, transmission, automatic, nzbget newsreader & rsync time machine backup.

Offline

 

#4 2010-03-31 01:28:06

switch10
Member
Registered: 2010-02-25
Posts: 27

Re: is my 323 killing hard drives?

Thanks for the replies guys.  Yeah I think I am done with RAID.  It seems much easier to just sync the disks with a cron job once a day.  I run Linux on all of my machines, and have set up many cron job backups with rsync, so this will be no problem.  There doesn't really seem to be any advantage to using RAID, other than it is easy to configure...

Offline

 

#5 2010-03-31 06:34:51

fordem
Member
Registered: 2007-01-26
Posts: 1938

Re: is my 323 killing hard drives?

switch10 wrote:

There doesn't really seem to be any advantage to using RAID, other than it is easy to configure...

The advantage to RAID1 is redundancy, and there's no way to achieve that redundancy without RAID.

Let's compare RAID1 (redundant mirrored disks) to what you are calling rsync mirrored disks, which in my opinion are not mirrored at all, but what I prefer to call scheduled synchronized disks.

Let's take a situation where you have a database holding the data for a real time transaction based system, stored on the primary disk in your scheduled synchronized array, and let's assume you're running rsync at hourly intervals.  Ten minutes after the last rsync job, the disk fails, not only have you just lost fifty minutes worth of transactions, but you no longer have access to the database and you're losing money by the minute until some one with the necessary skills remaps the database server to point to the correct disk.

Quite obviously, the amount of data lost will depend on how busy your real time transaction based system is and the time interval between rsync jobs.

Now - with RAID1 - where the disks are really mirrored - when one disk fails, all that happens is the system sends an alert to the administrator - no data is lost and more importantly the system continues to run allowing business to continue uninterrupted.  The administrator will then schedule the replacement of the failed disk at the most convenient time.

Now a RAID configuration may not be the ideal thing for you, but, it does have significant advantage and is used by big business everywhere.

Offline

 

#6 2010-03-31 08:50:58

switch10
Member
Registered: 2010-02-25
Posts: 27

Re: is my 323 killing hard drives?

fordem wrote:

switch10 wrote:

There doesn't really seem to be any advantage to using RAID, other than it is easy to configure...

The advantage to RAID1 is redundancy, and there's no way to achieve that redundancy without RAID.

Let's compare RAID1 (redundant mirrored disks) to what you are calling rsync mirrored disks, which in my opinion are not mirrored at all, but what I prefer to call scheduled synchronized disks.

Let's take a situation where you have a database holding the data for a real time transaction based system, stored on the primary disk in your scheduled synchronized array, and let's assume you're running rsync at hourly intervals.  Ten minutes after the last rsync job, the disk fails, not only have you just lost fifty minutes worth of transactions, but you no longer have access to the database and you're losing money by the minute until some one with the necessary skills remaps the database server to point to the correct disk.

Quite obviously, the amount of data lost will depend on how busy your real time transaction based system is and the time interval between rsync jobs.

Now - with RAID1 - where the disks are really mirrored - when one disk fails, all that happens is the system sends an alert to the administrator - no data is lost and more importantly the system continues to run allowing business to continue uninterrupted.  The administrator will then schedule the replacement of the failed disk at the most convenient time.

Now a RAID configuration may not be the ideal thing for you, but, it does have significant advantage and is used by big business everywhere.

Thanks for the reply. 

RAID has failed to rebuild twice on me already, so I already know it is not for me.

I do not understand why you say I could not read the data off of the second disk that is being backed up to.  Why would I not be able to access it? 

instead of mount //192.168.1.106/Volume_1 /media/drive
I would use mount //192.168.1.106/Volume_2 /media/drive right?

In my case this just seems to make more sense.  I am sick of RAID failing to rebuild itself..

Offline

 

#7 2010-03-31 14:47:04

fordem
Member
Registered: 2007-01-26
Posts: 1938

Re: is my 323 killing hard drives?

switch10 wrote:

I do not understand why you say I could not read the data off of the second disk that is being backed up to.  Why would I not be able to access it? 

instead of mount //192.168.1.106/Volume_1 /media/drive
I would use mount //192.168.1.106/Volume_2 /media/drive right?

In my case this just seems to make more sense.  I am sick of RAID failing to rebuild itself..

I'm not certain what data you are referring to here, but I'll try to clarify ...

I did say you'd lose fifty minutes worth of transactions - I got that backwards - ten minutes after the last rsync, you'd lose ten minutes worth of transactions, which is the data written to the first disk but not yet "synced" to the second - for the purpose of the discussion, it's not a major blunder, it's just intended to show that there is a window of opportunity for data loss - that your 'rsync mirrored disks" aren't really mirrored at all.

I also indicated you'd no longer have access to the database (and this is what I think you may be questioning) until some one with the necessary skills remaps the database server to point to the correct disk - the point is that with your scheduled synchronization, there is not a seamless transition if the primary disk fails - even though the data may be there, it is not accessible without operator intervention - until someone comes along and umounts the one disk and mounts the other (or remaps the drive letters) - with RAID, the need to do this is not there, the system does not see two disks at any time - it's one logical disk.

Offline

 

#8 2010-03-31 15:06:49

fordem
Member
Registered: 2007-01-26
Posts: 1938

Re: is my 323 killing hard drives?

switch10 wrote:

your drive might be fried.  I just had one do this, and so I pulled it, put it in a box and tried to format it, and it got to about 90% and started clicking.

I sent it back to seagate today..

I just found this in another thread...

http://dns323.kood.org/forum/post.php?t … ;qid=35665

Is this one of the occassions on which your RAID array failed to rebuild?  It sounds like a clear case of hardware failure to me, rather than a failure of the system or the RAID technology.

Failure of a drive during a RAID rebuild may cause the loss of data and the probability of this occuring has increased as drive capacity has increased - new RAID levels have been developed to mitigate the increased risk, and we now have RAID6 or as HP calls it, RAID with ADG (advanced data guarding), however, given the fact that the DNS-323 is a low cost, consumer NAS, these newer technologies are not offered.

It becomes a matter of deciding what's important to you - if the need for uninterrupted availability is the driving force, as it would be in a real time system, then RAID is the only solution that offers it,

Offline

 

#9 2010-03-31 15:51:46

obveron
Member
Registered: 2008-11-01
Posts: 36

Re: is my 323 killing hard drives?

for one thing, always make a raid 1 using two IDENTICAL disks, not just the same size, but the exact same LBA, and brand, rpm etc.

another the WD greens are not supported in raid operation, or for that matter, any platform OTHER than a windows machine.

Offline

 

#10 2010-03-31 18:06:55

fordem
Member
Registered: 2007-01-26
Posts: 1938

Re: is my 323 killing hard drives?

obveron wrote:

for one thing, always make a raid 1 using two IDENTICAL disks, not just the same size, but the exact same LBA, and brand, rpm etc.

This is a myth - it is standard industry practice to replace disks in a RAID array with not only different brands, but different capacities - in the past decade I have replaced so many failed RAID array disks with other brands & capacities that I have lost count.

Additionally, the RAID subsystem used in the DNS-323 is software (OS driven) RAID, the operating system treats the two disks as separate physical disks, but masks this from the user.

Offline

 

Board footer

Powered by PunBB
© Copyright 2002–2010 PunBB