Print

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

[harepal]

Hello,

I use a DNS 343 for 1 week without any problems.

But since yesterday i have a very huge lag (starting the NAS take 1 minute and half).
Same lag with web interface.

Lag stay with restart, shutdown and reset.

I install fun-plug to see what append to my NAS.

"top" indicates that [md0_raid5] take 70% of CPU and [md0_resync] 30%.

That's why i have lag.

But why i have this % ?

I saw this post http://forums.dlink.com/index.php?topic=12978.0

I have 2 Western Digital AV-GP S-ATA - 2000 Go - 32 Mo but i don't see anything about "Advanced Formatting Technology" in the data sheet.
The two others HD are Samsung Spinpoint F3 Ecogreen S-ATA - 2000 Go - 32 Mo

I try to rebuild the RAID, but with this lag Initializing don't make even 1% in 15 minutes.

After restart the NAS indicate "disk is full" because of the abort of rebuild, i can't acces NAS with telnet anymore.

If it's useful :
The "RAID 5 Sync Time Remaining" was between 1933.7 minute(s) and 4787.8 minute(s)

The only solution i see is to pull out the HD formatting them and pull them back.

Even if this solution work i lost all my data.

So what can i do to don't have this bug again ? (don't use RAID 5 anymore ? don't use HD western digital anymore ?).


PS : sorry for my poor English.

[dcmwai]

If you are able to telnet to the box again.

I would suggest to have a look on the raid with

Code: [Select]

# cat /proc/mdstat
That should tell us which disk was having issue.

If you are building RAID5 with 3 disk you wouldn't die if 1 disk fail.
That only happen if you have 2 disk fail at the same time....

Then ... cannot help...

That why when coming to buliding as raid 5 array...
Normally we would suggest to use hdd of different brand or manufecturing barch...

[harepal]

There is no fails in the RAID5 just a big lag (before the rebuild).

the 4 disks are working properly.

When a RAID5 have a disk damaged, he take 100% of CPU ?

What can explain this 100% ?

[dcmwai]

I think you have a more series issue...
IO Problem.

In linux system IO have to be sync.

In your case if one drive is physically slow in respond.. the whole raid group will be queue just for it...

using the smartd tools on the web to check which drive is more suspected to fail...

Get a new drive and remove the old one temporary to see if the the result will be better. **Power off 1st of course**

[ECF]

If you are using a WD10-EADS drive it is not and advanced format drive so thats not an issue. If your drives are as shown with a timer in the sync time remaining and does not say completed in the status page this is the cause of the lag. Have you run the smart test on all drives to check for a failing drive? I think this was all mention though.