Replaced smaller data hdd with larger - Have possible issue??

[flambot]

Greetings,

 

Read the wiki and relative forum pages as it's been quite a while since I upgraded a drive.

 

The wiki isn't absolutely clear on the exact procedure, but I followed some forum posts etc and things seemed to work. The main problem was the wiki never said anything about unassigning the old drive - or having to assign the replacement drive. Nor did it mention how long it would take the array to start once the "Start" array button was clicked. It took quite some time (many minutes) for the array to come on line. Now I'm a little confused that it says ParitySync / Data Rebuild. I think in V5 in just said "rebuild" next to the drive??

 

1 - Parity sync run previously

2 - Backup Flash

3 - Install new drive

4 - Start server

5 - Assign replacement drive

6 - Start Array

7 - Watch as drive rebuilds

 

Everything was looking great until it reached 0.6% - then the GUI stopped updating. The elapsed time is going up, but nothing out is increasing. Been sitting at 0.6% for over 15 minutes. Why??

 

Update: As I was typing this, the GUI suddenly updated and it is now at 1.5%. I'm new to 6.7.2 as I've only recently upgraded - so when weird things like this happen it makes me nervous.

 

Any insights??

 

Need some hand holding 😳

 

An observation...I got a message from CA saying there was something up (red box), but I'm confident is that there is a new version available (hence the result of the message). However, I noticed the CPU was about 60-70 utilized as long as I was watching - now it's only 25-30%. Very strange. I wonder if CA was doing something in the background to cause this? I noticed the GUI had responded again by the time I had noticed the CPU utilization was down.

 

 

[flambot]

When the rebuild is complete, should the parity sync after the rebuild be correcting or non-correcting?

[jpowell8672]

When the parity data rebuild finishes on the new drive you should be good but if for your peace of mind you wanted to run another parity sync then I would check the box write corrections to parity.

[flambot]

11 minutes ago, jpowell8672 said:

When the parity data rebuild finishes on the new drive you should be good but if for your peace of mind you wanted to run another parity sync then I would check the box write corrections to parity.

Thanks muchly. This was the way I was leaning.

[John_M]

@flambot

You mean a parity check, not a parity sync. It should be a non-correcting one. Here's why: if your disk replacement went well there should be no error in parity, so any discrepancy revealed by the parity check would mean that the rebuild went wrong somewhere. If your parity check is a non-correcting one and it reveals errors you can still have another go at rebuilding the data disk. If your parity check is a correcting one and it reveals errors all you are doing is making parity agree with the bad data disk and ruining any chance of recovery.

[itimpi]

2 hours ago, flambot said:

When the rebuild is complete, should the parity sync after the rebuild be correcting or non-correcting?

If the rebuild was error free then parity should be correct so I would think that non-correcting was better!

[flambot]

Thanks for this. That makes a lot of sense. So non-correcting it is. I've only started recently doing non-correcting as that has always been the default of unRAID. However, after reading many posts on the subject, I figured a non-correcting one was at times good practice. I was thinking that the new writes to the new disk needed to be checked (after the fact) to confirm that the writes were all successful. However, my thinking is probably suspect as I certainly don't know a lot about these things. If I think on it further, I guess unRAID will somehow tell me of failed writes - probably in the GUI??. I have never had one (touch wood) so I'm not sure of the process.

 

@John_M -

Quote

You mean a parity check, not a parity sync.

It is only as you've pointed this out that I realised I always thought they were the same thing. However, this is now clear to me they are not. Sometimes I realise how thick I really am! 😵

 

My rebuild is progressing - currently at 17.1% after 4h51m.

 

Once this is done, I plan to copy data from a small drive, then rebuild that small drive to a larger drive using the XFS file system. This is a first for me - and even scarier than rebuilding a drive!! 😬

 

Really appreciate the input.

[jpowell8672]

1 hour ago, itimpi said:

If the rebuild was error free then parity should be correct so I would think that non-correcting was better!

Ok now I am a bit confused. He did a pre parity check and if there was no errors then parity is correct. He rebuilds the drive and if there are no errors then parity is correct. So then if he does a non correcting parity check and there is errors why would that be a bad rebuild when it reported no errors when completed plus previous parity check was correct. Are you saying during rebuiild it also rebuilds/replaces the parity? If he did a correcting parity check and there were errors wouldn't it just fix them rather than rebuilding the drive again?

[jpowell8672]

Ok I have a better understanding of how parity works now after reading:

https://wiki.unraid.net/Parity

 

[itimpi]

2 hours ago, jpowell8672 said:

Ok now I am a bit confused. He did a pre parity check and if there was no errors then parity is correct. He rebuilds the drive and if there are no errors then parity is correct. So then if he does a non correcting parity check and there is errors why would that be a bad rebuild when it reported no errors when completed plus previous parity check was correct. Are you saying during rebuiild it also rebuilds/replaces the parity? If he did a correcting parity check and there were errors wouldn't it just fix them rather than rebuilding the drive again?

During a rebuild the system is rewriting the replaced drive so that it’s contents agree with what the combination of the other drives plus the parity drive say it should be.   UnRaid will detect the case of the disk drive reporting a write error, but what it will NOT detect is the case of the write appearing to be error free but a subsequent read returning the wrong data.    In theory this should never happen, but it is always a possibility if the new drive is faulty in any way.     Doing a parity check after the rebuild is just a confidence check that this (remote) possibility has not happened.

[flambot]

The drive rebuild without incident and completed after 25 hours. The next step is the copy the data from my next small drive so it is free to upgrade to xfs.