I am running 3 Plextor NVMe M8Pe(Y) 4X devices 1TB in size configured as one volume using spanned devices. I am trying to offload the device to a rotating storage device. It runs anywhere from 1 hour to 6 hours and then I get stornvme into the picture to reset the device and at that point it drops out. One of the devices are installed on the root bus and the other two are on a port with a bridge. It almost looks like stornvme is resetting the bus to shut down the I/O operations and whacking the other device on the bridge but not restoring the configuration space for that one. It takes a reboot of the workstation system to get the device back on line again.
Is anyone familiar with the operation of stornvme that could explain what it is doing? I have a lot of experience with PCIe Reset from dealing with it on Linux systems in my former job and it looks like a clear is being done at the wrong level of the Bus layers and the config is not being restored properly. I have also seen some "Surprise down" errors reported in the event log which means that a device was taken off the bus without the proper suppress bits for power loss errors due to MC control functions.
Thank you in advance.
Is anyone familiar with the operation of stornvme that could explain what it is doing? I have a lot of experience with PCIe Reset from dealing with it on Linux systems in my former job and it looks like a clear is being done at the wrong level of the Bus layers and the config is not being restored properly. I have also seen some "Surprise down" errors reported in the event log which means that a device was taken off the bus without the proper suppress bits for power loss errors due to MC control functions.
Thank you in advance.