Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[cawli2]

I had the misfortune of losing a drive in my DNS-323 and had to replace it and rebuild the RAID-1 mirror.

These are 1000Gb drives.  It has been syncing for over 17 hours, and the web page displays a "Sync Time Remaining" value that appears to be random.  I expect it to countdown until completed, but what I see are values in the thousands or hundreds, completely random upon each refresh of the screen.

I can see my data, so the 'master drive' is good.  I am running the latest firmware 1.10.

I decided to do an orderly restart.  The GUI now displays 14280.8 minutes.  Complete nonsense.  I've seen numbers as low as 198 minutes, but that was just a meaningless random number like all the rest.  Clearly a bug.

However, the question remains - IS it actually rebuilding the mirror, and HOW LONG should it take for a 1tb drive?

Wayne

[cawli2]

I have an update to this problem ... still not finished synching, but I noticed that a process mt-daapd was using 80+ % of the cpu (ran a top command from an ssh session).  This is a media server program .... so I disabled the iTunes server, and now the top cpu users are md0_resync and md0_raid1.

Now that's more like it.  Why on earth a media server would take precedence over a RAIDset synch beats me.

Further, the sync time remaining has now settled into a much better ~120 minutes, but oscillates within a much tighter range - +/- 20 minutes.  Still, the time remaining appears to be poorly estimated and until I see some significant reduction in this time, it is still random as far as I'm concerned.

Perhaps it will oscillate within a tight range with an overall reducing trend.  I can accept that as a 'best estimate' provided the trend continues downward over time.

Wayne

[cawli2]

Success!  The countdown from ~120 minutes, with all its oscillations +/- 20 minutes, eventually did count down and complete.

SO ... disable your iTunes and UPnP AV Server and anything else that might be interfering with the rebuilding process.

(I was able to use ssh to connect and see what was going on.  I use fun-plug for this, and as a side note, I had to disable fun-plug in order for the device to format the new drive properly.  Once formatted I re-enabled fun-plug so that I could ssh into the machine once again.)

I cringe when I think that there are non-IT professionals out there using this device.  D-Link needs to 'user-proof' their NAS devices.  Typical end-users are not going to hack their DSN-323 with fun-plug and know enough about Linux to see what is happening on their NAS to address problems such as I encountered.

Wayne

[JavaLawyer]

Thanks for posting your troubleshooting. Enabling too many peripheral services significantly increases overhead on many of the legacy ShareCenter models, sometimes at the cost of performance and slowdown of other critical services.

[fordem]

It is the norm for raid rebuild to have a low, if not the lowest, priority - RAID is designed to to have the data always available for use.