If began to do a few changes recently, in particular, upgrade to UnRaid 5.0.5. and now upgrade my Parity from a 2TB to WD 4TB Red drive.  I did pre-clear the drive before adding it to the array, which took something like 60 hours.  I added it in the former parity drive slot (as my server has no expansion bays left - I actually precleared the drive sitting it beside the case on a box).

UnRaid did its automatic process and copied everything necessary so that my array is protected again.

 

The question:  Is their a need to run a Parity Check after just writing all the information to the new Parity Drive, or is their somewhat of an automatic check that is done while the data is written for the first time?  I've read one person's suggestion, from a few years ago, which was "yes", to running it, but the reason was just as a greater burn-in test (sounded like he may have had a drive in the past that wasn't D.O.A., but was "Dead within the week".

 

After this, I can re-task the old Parity to replace my smallest drive (750GB).  Not a giant gain, but a gain nonetheless.  Trying to postpone the upgrade to a Norco as long as I can! 

Yes - it is a good idea (albeit not critical) to do a non-correcting parity check after calculating parity for the first time as a confidence check. 

 

This will check that the data disks return the same values to the read from parity check, and also that the read of the parity disk returns the expected value.

Yes - it is a good idea (albeit not critical) to do a non-correcting parity check after calculating parity for the first time as a confidence check. 

 

This will check that the data disks return the same values to the read from parity check, and also that the read of the parity disk returns the expected value.

 

I would definitely +1 that .... how do you know that the parity itself has been successully written unless you do a check (non-correcting as suggested)?

So, I ran the parity check (well, its 600GB in to it).  I just realized you guys said "Non-correcting" parity check.  I missed the "non-correcting" check box.  Assuming I check the output and/or syslog and it makes no reference to any corrections being made, then I take that to mean "all is well".  Reasonable conclusion, based on me missing that check box and if the output is clean?

 

I could stop the parity check and start it over again, but I'd rather not if not needed.  Moving to the 4TB hard drive has its drawbacks - mainly being that it takes forever to preclear and run a parity check on due to the large number of sectors.  With respect to the time it takes to complete, is that mainly a factor of the HDD speed or does my $50 CPU play a roll in this?  During my original build 5 years ago, or so, CPU wasn't important, as it never really did much (at least for my needs, which were simply to run UnRaid with no real add-ons), so I just got a Celeron CPU for around $50.

No need to stop the check.  In fact, if you ran a non-correcting check and it found errors, you'd then want to run a correcting check to fix them (i.e. after a parity sync there's no reason to do a non-correcting check ... just do a standard correcting one).    After a drive rebuild, it's a good idea to do a non-correcting check; as if something went wrong with the rebuild you'd want to do it again, and wouldn't want parity to be altered.    But in most cases, a standard correcting check is what you want to do.

 

"Last checked on Mon Jun 9 01:35:19 2014 EDT, finding 0 errors."

 

Looks like I am good to go, with the new parity drive!

 

I'll put it through another test tonight, as my next step is to replace my smallest drive (750GB) with the former parity drive (2TB).  Just deciding if I should preclear the old parity drive or not.  I've got 11 hours before I get home to decide that!

 

Thanks for the info/opinions to all those that replied! 

I'll put it through another test tonight, as my next step is to replace my smallest drive (750GB) with the former parity drive (2TB).  Just deciding if I should preclear the old parity drive or not.  I've got 11 hours before I get home to decide that!

As long as you have a current smart report on the 2TB and it looks good, there is no need to preclear it. Preclear doesn't effect rebuild times when you are replacing a drive, and it's already been thoroughly tested by use as the parity drive, so I say check the smart report and start the replacement process. Just don't blow away the 750GB until the rebuild is successful. Many times when I do a voluntary drive replacement, I will do a file binary compare on the old and new drives to satisfy my need to know if the array is doing its job. I've never had a miscompare, so I guess it's a waste of time, but it makes me feel better.

Agree -- keep the old drive handy just in case.    When the rebuild is done, you can verify it in one of two ways:

 

(1)  Attach the old drive to a PC and run a comparison utility to confirm the contents of that drive and the new drive are identical;

 

or

 

(2)  Run a non-correcting parity check.    This is the one time I suggest using the non-correct option; as if it finds any issues the rebuild may not have been successful.    I've never seen a case like that ... but it CAN happen.

 

I like Gary's suggestion, for the simple reason I have no idea how to do a "file binary compare" 

 

Gary's makes sense.  A non-correcting parity check, after the rebuild, would ensure that every sector has the proper value.  I'm guessing it make take a little longer to run another full parity check, especially having moved up to the 4TB world, but its just time, nothing new to figure out.

 

I also agree with the part about not re-tasking the smaller 750GB drive I am removing, until after I know everything is all good with the rebuild on to the new drive.  I have no immediate need for it anyways, so it can collect dust for a little while 

Everything has gone well.  I now have 1.25TB of free space to work with, however on my main computer, I have even more ripped content to transfer over.  I just started the first bid data transfer, through a gigabit router the pc and server both connect to.  Both machines have gigabit ethernet ports.  I am moving 684GB of mainly 4GB TO 7GB ISO rips.  The estimated time to complete is nearly 24 hours.  This my first major transfer from outside the server since moving to the WD 4TB RED Parity drive.  Im thinking the transfer speed is slower than before, but i may be wrong.  I am not use to moving this much data.

 

Does that time frame seem reasonable , bearing in mind I do not use a cache drive?

 

Thanks 

That's not as quick as I'd expect, but Windows is notoriously inaccurate at estimating the time to completion -- so it'll likely be less.  For that much data I'd expect 6-8 hours.  What does it show for the transfer rate?    If you're not getting above 10-11MB/s then your network is likely operating at 100Mb speed instead of Gb.

 

If you are using Windows suggest you install teracopy. It has an option to verify the copy, and pretty accurately estimates the total time. It also has other convenient features, like waiting for one copy to complete before starting the next, and the ability to pause a copy operation.

 

At 30MB/sec, 700G should take about 6.5 hours. More in verify mode.

Last night, i believe it was only running around 10 MBps , however thanks to good old Windows and its "Critical Update", it rebooted my main PC at some random time in the night.  I started up moving some smaller batches of files this morning and the speeds look more normal, at about 23 MBps .

 

bjp999  - thanks for tip.  I will check that out!