Winstep

Software Technologies


 Winstep Forums


Print view
Board index : Winstep Forums : Articles  [ 93 posts ] Go to page Previous  1 ... 3, 4, 5, 6, 7
Author Message
 Post subject: Re: New system is here: Intel Optane 905p inside!
PostPosted: Sat Feb 15, 2020 5:47 pm 
Offline
Site Admin
User avatar

Joined: Thu Feb 26, 2004 8:30 pm
Posts: 10823
Currently the system has a 980GB 905p Intel Optane drive, a 2TB Samsung 970 EVO NVMe drive and a Crucial MX500 1TB SATA SSD.

Every week a full backup of the data in all these drives is automatically made by Acronis True Image to an external WD 6TB drive connected via USB 3.0. Every other day except Sundays, a differential backup based on this full backup is also made. Every time a full backup is performed, the previous full backup is renamed to .old.

Likewise, every 4 weeks on a Sunday a full backup of all the drives is made to another WD 6TB external drive. On the other three Sundays a differential backup based on this full backup is made. Every time a monthly full backup is made the previous full backup is also renamed to .old.

This backup scheme, which I have described previously in other topics, ensures that I have both redundancy (in case one of the two external hard drives fails) *and* also a way to 'time travel' by up to two months into an old backup in case a file gets accidentally deleted and I don't notice it immediately.

Usually it takes about 3 and a half hours to backup the over 2TB of data (compressed) spread out among the internal drives, which is never a problem even when I am working as the backup happens in the background.

Sometime around the middle of January I noticed the full backup had taken an extraordinary amount of time to complete, something like one and a half to two hours longer than it normally does. This prompted me to investigate.

What I found out by looking at the Windows Event Log was that the 2TB Samsung 970 EVO - which is barely over 1 year old and only has about 3.38 TB data written into it so far) had thrown a disk read error. In fact, SMART data for this drive reported 3 Media Errors (the HDD equivalent of a bad block/sector).

Now, NAND flash memory, just like the magnetic sectors in a hard disk, can also go bad. Usually the drive controller will remap the bad NAND cells and replace the affected area with good NAND cells taken from the spare area that all SSD drives have.

I wasn't too thrilled with the 3 bad blocks in a basically brand new NVMe drive, but it's not alarming until bad blocks actually start pilling up. The backup also completed, which in principle meant that the 970 EVO drive had eventually managed to retrieve the correct data from the faulty NAND cell(s) after retrying for a while.

The next full backup I was watching what happened like a hawk, but it went by with no issues, so I started to relax a bit. Alas... after remaining at 3 media errors for nearly two weeks, the drive eventually threw a 4th media error while being backed up. Again the backup managed to complete, but now I had alarm bells going off in my head all the time: despite having backups, there was a chance data was actually getting corrupted.

Fast forward to last week and the drive throws 2 more media errors. This time the backup aborts with an error too. Ok, this drive can no longer be trusted and it seems to be failing fast - time to get a RMA on it from Samsung.

So I sent an email to Samsung support explaining the situation and giving them all the details I could. At the same time I was apprehensive given that Samsung Magician still reported the drive as 'Healthy' despite the ongoing media errors. Given this Samsung could simply refuse to honor the warranty and I would be stuck with a 600 Euro NVMe drive that I could no longer trust.

Their response was:

Quote:
As a first step towards initiating the RMA procedure for your drive, we will need the following information, if possible:

Please provide us with screenshots of any error messages or error codes.

1. A screenshot of the Samsung Magician Drive Details page
2. The SMART test result from Samsung magician.
3. A copy of your proof of purchase
4. What is your country of residence?
5. Photos of the front and backside of the entire SSD.
6. Are you an end-user or re-seller of this drive?

We will review the information and advise you accordingly in regards to RMA.


So, I sent them all the information they requested *except* the screenshots of the drive. I didn't really want to remove the drive until I was sure I could get an RMA based on the SMART data I sent them, etc...

Either way I overnighted a new 2TB 970 EVO Plus from Amazon.it - I would have to get a new drive anyway since I would have to have something to replace the old EVO while it was on its way to Samsung.

Attachment:
20200214_114833.jpg
20200214_114833.jpg [ 1.67 MiB | Viewed 862 times ]


Good thing I did too since Samsung were adamant about the screenshots of the drive BEFORE issuing a RMA (which pissed me off a bit, although I understand why they want this).

So when the new drive arrived I removed the NVMe heatsink on the motherboard, installed the new drive (this motherboard supports up to two NVMe drives), copied all the data over from the old 970 EVO to the new 970 EVO Plus, removed the old 970 EVO and took the screenshots they asked for. A few hours later I got the RMA from Samsung.

Hopefully they will send me a *new* 970 EVO back, and despite now having one more NVMe drive than I really need, at least I won't be also stuck with a failing NVMe drive (or a drive that cannot be trusted, which is the same) that cost me over 600 Euros a little over a year ago (the new 970 EVO Plus is not only a bit faster as it is also cheaper, it 'only' cost 450 Euros).

Something else that pissed me off is that there is no way for the end user to test the NAND cells on the EVO drives, like you can check the surface of hard disks for bad blocks with the CHKDSK /R command. Even the 'Diagnostic Test' option in Samsung Magician is disabled for these drives.

###

The heatsink on the motherboard, below which you install the NVMe drives:

Attachment:
20200214_120157.jpg
20200214_120157.jpg [ 2.35 MiB | Viewed 862 times ]


The two drives in place without the heatsink and with the Intel Optane temporarily removed for 'elbow space':

Attachment:
20200214_121229.jpg
20200214_121229.jpg [ 2.55 MiB | Viewed 862 times ]


The Plus version of the 970 EVO is indeed a little bit faster (as I already expected it to be given the reviews) but it also seems to run a few degrees hotter (although not enough for it to be a problem, when transferring files between both drives it didn't go above 71C for the NAND and 75C for the controller - contrast that with the older EVO's 61C for the NAND and 70C for the controller). Might also have something to do with which half of the IO shield/heatsink the drive is installed under, I will have to test that later.

_________________
Jorge Coelho
Winstep Xtreme - Xtreme Power!
http://www.winstep.net - Winstep Software Technologies


Back to top
 Profile WWW 
 
 Post subject: Re: New system is here: Intel Optane 905p inside!
PostPosted: Wed Mar 18, 2020 4:47 pm 
Offline
Site Admin
User avatar

Joined: Thu Feb 26, 2004 8:30 pm
Posts: 10823
Ok, Samsung just sent me a **brand new*** 2TB 970 EVO **Plus*** to replace my (older) 2TB 970 EVO. Not bad. :)

_________________
Jorge Coelho
Winstep Xtreme - Xtreme Power!
http://www.winstep.net - Winstep Software Technologies


Back to top
 Profile WWW 
 
 Post subject: Re: New system is here: Intel Optane 905p inside!
PostPosted: Sat Mar 21, 2020 6:47 pm 
Offline
Site Admin
User avatar

Joined: Thu Feb 26, 2004 8:30 pm
Posts: 10823
So, interesting thing: now that I have two EVO Plus, I can confirm that the Plus actually runs warmer than the previous EVO version.

The old EVO ran at 35ºC for the NAND and 38ºC for the controller when idle. The EVO Plus is actually the other way around, the NAND is kept at a higher temperature than the controller: 42ºC for the NAND, 39ºC for the controller.

Which makes sense since NAND actually needs to run warm to operate reliably. Perhaps that is why my previous EVO ended up developing bad blocks, as this PC case is optimized for airflow.

Anyway, with the new NVMe drive, I now have a mind boggling total of 6 TB of **flash** storage in this system (1TB Intel Optane + 4TB 2 x 970 Evo Plus + 1TB Crucial MX500)

_________________
Jorge Coelho
Winstep Xtreme - Xtreme Power!
http://www.winstep.net - Winstep Software Technologies


Back to top
 Profile WWW 
 
Post new topic Reply to topic Board index : Winstep Forums : Articles  [ 93 posts ] Go to page Previous  1 ... 3, 4, 5, 6, 7
Display posts from previous:  Sort by  

Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to: