I have set up a VM runing Win8.1 on the cache drive. In addition, this cache drive also holds some files. When I have mapped the cache drive as well as some array drives within Win8.1 (VM).

 

When copying from my cache drive to an array drive, I am getting the following error message:

 

" An error is keeping you from moving the file. If you continue to receive this error, you can use the error code to search for help with this problem. Error 0x8007003B: An unexpected network error occurred *

 

Any thoughts appreciated. Thanks!!!

The same error also occasonally appear when copying files from insides the VM to the array. In general, any copying frmo within the VM (i.e., one array disk to another array disk triggered from the VM) is very buggy. Is this expected behavior and I should not use the VM (Win8.1) for copying files?

Any thoughts?

 

Basically, any copy activity triggered from the VM either stops with an error or is incredibly slow. This includes VM to array and also array to array (when initiated from VM),

 

As context, my VM runs on a cache drive, which is a SSD on PCI-E.

 

Thanks in advance for any help!!!

Any thoughts?

 

Basically, any copy activity triggered from the VM either stops with an error or is incredibly slow. This includes VM to array and also array to array (when initiated from VM),

 

As context, my VM runs on a cache drive, which is a SSD on PCI-E.

 

Thanks in advance for any help!!!

 

So the error you are getting is from within the Windows VMs themselves, yes?  I have not had any issues copying files in the scenario you describe, and my performance has been just fine.  I searched google for Error 0x8007003B: An unexpected network error occurred and found some results on MS and Synology's website that points to this being an issue outside of the NAS itself (either the router or something in Windows).

Thanks. My hypothesis is that the issue relates from my SSD PCI-E card, which is a rather fancy, while unique setup. I am trying to move my VM to a "normal" SATA drive and see whether the issue remains.

I tried and copying from a regular SATA disk (cache drive) works. So, the issue indeed is related to my PCI-E SSD. That's a pity as I really love this device (and it gives me a 17th disk). Any idea whether this could be a driver issue or what else I could try to make it work?

I tried and copying from a regular SATA disk (cache drive) works. So, the issue indeed is related to my PCI-E SSD. That's a pity as I really love this device (and it gives me a 17th disk). Any idea whether this could be a driver issue or what else I could try to make it work?

Could be driver related. Your test certainly points to that, but its still odd.  Can you link to the mfg website for that device so we can get exact make and model info? 

It is very easy to replicate. Copying now (from SATA disk) works very well in all combination. When I used to run my VM from the PCI-E SSD, copying was basically not possible when "triggered" from the PCI-E SSD drive (neither from VM to cache, not from VM to array, nor from array to array).

 

Please see below link to my device:

 

http://www.newegg.com/Product/Product.aspx?Item=N82E16820249043

Any thoughts on it? In the meantime, I stopped using it and went with a "normal" SATA disk as cache drive. However, I really like my super-fast PCI-E SSD and it also saves me a SATA port that I can use for data storage instead. Any thoughts whether there is a chance to get this working?

Any thoughts on it? In the meantime, I stopped using it and went with a "normal" SATA disk as cache drive. However, I really like my super-fast PCI-E SSD and it also saves me a SATA port that I can use for data storage instead. Any thoughts whether there is a chance to get this working?

 

Whoops, forgot to check this one out again.  Would you mind running the following command on your system:

 

lspci -k

 

Copy and paste the output back here.  I want to see the device information for the PCI-E SSD and what driver is being loaded for it.

Any thoughts on it? In the meantime, I stopped using it and went with a "normal" SATA disk as cache drive. However, I really like my super-fast PCI-E SSD and it also saves me a SATA port that I can use for data storage instead. Any thoughts whether there is a chance to get this working?

 

Whoops, forgot to check this one out again.  Would you mind running the following command on your system:

 

lspci -k

 

Copy and paste the output back here.  I want to see the device information for the PCI-E SSD and what driver is being loaded for it.

 

Thanks, this is great. Do I need to set it up as cache drive for this test. I realized from previous trials that just changing cache drives for trying out things does actually not work and gets me into trouble. Any chance to get the info without re-assigning my cache drive?

Any thoughts on it? In the meantime, I stopped using it and went with a "normal" SATA disk as cache drive. However, I really like my super-fast PCI-E SSD and it also saves me a SATA port that I can use for data storage instead. Any thoughts whether there is a chance to get this working?

 

Whoops, forgot to check this one out again.  Would you mind running the following command on your system:

 

lspci -k

 

Copy and paste the output back here.  I want to see the device information for the PCI-E SSD and what driver is being loaded for it.

 

Thanks, this is great. Do I need to set it up as cache drive for this test. I realized from previous trials that just changing cache drives for trying out things does actually not work and gets me into trouble. Any chance to get the info without re-assigning my cache drive?

 

No need to change anything on your system.  I just need you to have the PCIE SSD physically plugged in on the system when you run the command.

Cool, thanks! Please find driver info below:

 

SATA Controller: Lite-On IT Corp. / Pextor M6e PCI Express SSD [Marvell 88SS9183] (rev 14)

Subsystem: Marvell Technology Group Ltd. Device 9183

Kernel driver in use: ahci

Kernel modules: ahci

 

Thanks in advance!

Cool, thanks! Please find driver info below:

 

SATA Controller: Lite-On IT Corp. / Pextor M6e PCI Express SSD [Marvell 88SS9183] (rev 14)

Subsystem: Marvell Technology Group Ltd. Device 9183

Kernel driver in use: ahci

Kernel modules: ahci

 

Thanks in advance!

Ok, this could be because it is getting a standard driver, not one that is meant for the PCIE SSD. I will have to look and see what Linux kernel drivers exist for this device and see about incorporating them into the next release.

Ok, this could be because it is getting a standard driver, not one that is meant for the PCIE SSD. I will have to look and see what Linux kernel drivers exist for this device and see about incorporating them into the next release.

 

Thanks for looking into this, very much appreciated!!! Happy to test if you have some testing code with a new driver.

 

 

Tom and I were talking on this earlier today and we looked into your specific device, hoping that it would have support for NVMe, but it doesn't appear to, which means this isn't a driver issue.

 

I'd like you to test installing a Linux VM and attempting the same thing if you could (copying data using the VM from the cache to the array).  Any chance you could do that for us (using your PCIE SSD)?  You can stop the array, unassign your current cache drive, then assign the PCIE SSD to the cache, start the array, conduct testing, then when done, stop array, and reverse.  The cache, unlike array devices, doesn't require you to do a "new config" to remove / add devices from it.

Thanks for looking into this. Very much appreciated.

 

I am a bit hesitant to change the cache drive as I had some issues with this in the past (http://lime-technology.com/forum/index.php?topic=42650.0). Any idea what I did wrong before? After unassigning the cache disk, I could not re-assign later? I can give it another try, but would be great if you can take a look what I did wrong before.

 

Would it do the same trick if I do a "new config" (which worked without issues in the past). I could add the PCI-E drive to the array and try copying from it. Also, do you think it could be formatting issue (XFS vs. btrfs)?

I did some more tests. I added the PCI-E drive the array (formatted as XFS and then later also as btrfs). This works well and speed is great.

 

Next step is to try how the speed is when mounted as cache disk. If yet again good, I will give it a shot to install a Linux VM. Is it worth to try Win8.1 first? When it was not working, the VM was running Win7?

Speed is also ok when mounted as cache disk. Next step is the Linux VM. No clue how to do this though. Any step-by-step guide?

It is working now. No idea what is different, but things work well now (touch wood).

 

A few things changed since my initial attempt (Win10 instead of Win7, upgraded to Unraid 6.1, allocated 12GB Ram instead of 6GB to the VM). No idea what else is different, but good news is that it is working now and I can copy from the VM to the array at reasonable speeds.

It is working now. No idea what is different, but things work well now (touch wood).

 

A few things changed since my initial attempt (Win10 instead of Win7, upgraded to Unraid 6.1, allocated 12GB Ram instead of 6GB to the VM). No idea what else is different, but good news is that it is working now and I can copy from the VM to the array at reasonable speeds.

Hmm, that's interesting. Keep me posted as testing continues.

Agree, it is very weird and I doubt any of the changes can explain it. I wish it would, but I am concerned that some form of error may surface again.

 

Somewhat, I feel that my Unraid server is still not too stable. For example, I had an I/O error with one of the disks a few days ago (which then went away overnight). Also, occasionally, disk unmounted, but then are back after rebooting.

 

Also, when copying from array disk XXX to array disk YYY works, but at the same time I can not copy from disk ZZZ to AAA (and even read.activity is very slow). Mich slower / more sluggish compared to when I used Win7 natively rather than over VM. But I can live with this and it is probably caused by the nature of virtualization.

 

One other thought is that I may be pushing Unraid to its limits with 17 disks (one cache, one parity and 15 array disks) for a total of around 80TB. Maybe, that's more than what it was initially designed for?

One other thought is that I may be pushing Unraid to its limits with 17 disks (one cache, one parity and 15 array disks) for a total of around 80TB. Maybe, that's more than what it was initially designed for?

I do not think that is excessive.

 

There have certainly been reports in the forum those who have reached the limit of 25 devices allowed by the Pro license, and also have had extra disks mounted outside of the unRAID array.

I changed my setup and unfortunately I am facing the same issue again as I did two years. Never change a running system...

 

Anyhow, it is done now and I have a new mobo (Asus X299) that supports two M.2 devices. I have installed my "old" Plextor M6E and also added a Samsung PM961 M.2 NVMe PCIe SSD. I moved my VM to the new Samsung drive.

 

Unfortunately, the above mentioned error is back (Error 0x8007003B). I have no clue how it was fixed two years ago, so am back to this thread.

 

I ran lspci -k as advised before. Please see below results. Any thoughts appreciated:

 

https://pastebin.com/jsG6Ukk3