microsoft / StorScore

A test framework to evaluate SSDs and HDDs
http://aka.ms/storscore
MIT License
81 stars 34 forks source link

StorScore Corners recipe creates disk corruption messages in eventvwr during purge phase #50

Open dsgulla opened 7 years ago

dsgulla commented 7 years ago

With the latest June StorScore version, I am seeing an issue on Windows 2016 server edition where the storscore Corners.rcp file is causing drive corruption messages to pop up in the eventvwr. I am using a drive that supports Secure Erase. I see this issue if the storscore partition is already created and when it runs the "secure erase/ diskpart clean" during the purge phase. wanted to check if its a known issue.

Thanks Deepak

lauracaulfield commented 7 years ago

Hi Deepak,

Thanks for reporting this. This behavior is new to me. The June release with Windows Server 2016 are the first version of StorScore and Windows that support sending secure erase to the drive. (So you may be uncovering a latent issue in the drive or OS.)

Does StorScore state that the secure erase operation completed successfully? The NVMe command that is sent is the "Format NVM" command with the secure erase bit set. Do you know if the drive implements this command correctly?

-Laura

dsgulla commented 7 years ago

Thanks for the response Laura, yes the crypto erase is successfull. I have tried to erase it via storagetool separately aswell and see a successfull message and also no issues seen from our drive. But when there is a partition/FS and during storscore purge phase, I am not sure exactly the sequence of event. When I look at my drive, I do see the message popping up in the eventvwr around the time when crypto erase is being issues which corelates to the timing.

Attaching a few snapshots.

image

image

PS C:\Software\StorScore\bin> .\StorageTool.exe -Secureerase disk 1 Disk #1 : [NVME ] [SSD] [01 00 00 00] Secure erase was successful PS C:\Software\StorScore\bin>

I have verified that this issue is mostly seen with Corners recipe, on windows 2016 Server and with any Enterprise grade NVMe SSD.

However, If I just issue a secure Erase by itself, I dont see this corruption message being generated. Thanks Deepak

lauracaulfield commented 7 years ago

Interesting -- thanks for the extra info. Is the issue correlated with the corners recipe or with the existing partition? Have you tried running the storage tool on the drive when it has a partition and file on it?

The sequence of purge and disk preparation starts in lib\target.pm around line 350. The first step (if it's supported) is to secure erase the drive. Then diskpart "cleans" the drive, creates a partition, and creates a file. Is it possible the secure erase doesn't complete within the drive before the host creates the file system (and then the remaining work of secure erase deletes something that was later written to establish the file system)?

dsgulla commented 7 years ago

I have tried to SecureErase when the FS/Partition is on but I dont see the issue. It is something that is being done during purge phase,(Probably what you explained could be happening), where the secure erase may not be completing buy the time the FS is being created.

but here is an easy way to recreate it. I have also verified that its not specific to Corners recipe. It occurs with other recipe files aswell.

steps:

Thanks Deepak

dsgulla commented 7 years ago

Hi Laura, Any update on this issue. All tests are consistently failing.

Thanks Deepak