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

Thank you for the reply jesbo. 

ok, let me rephrase...

Can anyone honestly state that they had a drive fail while in Raid 1 in the DNS-323, and they merely powered down the unit, then ejected the bad drive, then replaced with a new drive, powered back up, and it actually rebuilt the mirror with NO DATA LOST??? That is the question. Does it do what it's supposed to????

If it doesnt, then why are people(including myself) still using the thing(Russian roulette), other than as a toy to run linux scripts on to test theoretical "thats kewl" scenarios??? I mean, I'm afraid to trust it now that I really need a DEPENDABLE unit that functions as designed.

Last edited by osxtasy (2008-11-25 04:44:02)

+1 to mig.

I liked the DNS-323 for the potential of reducing the number/frequency of my home data backups.  However, after more than 30 years in IT I know only too well that nothing is fool proof - not even the high end storage systems costing many thousands of $$.  So I back it up just as I would any data on storage media - for 2 reasons: Disaster recovery, and point in time data recovery.  There are occasionally times when I screw up data on my own, or inadvertently delete and need to recover a file or group of files to a point in time (usually the last backup).  I still religiously do backups to offline media.  Perhaps a bit old-school, but it works for me.

I agree that a full set of faulure test scenarios for the DNS is in order (even for me) and you must understand how to recover from them.  The wrong time to try to figure it out is AFTER the drive fails with live data on it and no offline backups.

When you do your failure scenarios, by all means document your process and findings.  I will do the same.  I may not get around to doing it, though until after the holidays.

My thinking for a disaster test is to power off the unit, remove a drive, install the removed drive in a spare windows machine and delete all partitions.  Then verify that the DNS is still functional with one working drive and that data is accessible.  Then power it off, reinsert the blanked disk and go from there.  This should simulate the installation of a new disk and allow for Raid 1 mirror rebuild.

If this is successful, repeat the above process with the second disk.

luusac wrote:

But the 323 stores the drives serial # in flash - testing it this way wouldn't be the same as putting in a different drive (which would have a different serial), would it?  You can of course update the serial in flash ...

Thanks for the heads up on that.  I would think the RAID software should be able to handle the situation that it sees the same serial number disk with missing partitions and still be able to rebuild the mirror.  But true - its not the same failure scenario as a true drive failure with a different physical disk replaced.  I guess my failure test scenario is focused more on seeing it rebuild the disk including setting up the correct partitions, file systems, adding it to the metadisk and syncing the data.  If it can do that on the same disk, I would expect it to do it on a brand new disk (different serial number) as well.

Keep in mind, the serial number storage and config files are a function of the DLink F/W, not the RAID software.  The RAID software works fine, the bugs appear in the DLink side of things.