Hi all, I'm new around here, so I appreciate any help and debugging tips with a problem that has been driving me crazy for the past week, ever since moving my Unraid instance over to a new computer.

 

My diagnostics are attached.

 

Since switching to a new computer, my unraid instance has been having issues with the USB flash drive, only when I'm running my instance of Frigate. With Frigate off, the server runs for days with no problem, but after starting Frigate, the server runs for an hour or two, and then shows the `USB Device is corrupt or unavailable` error. I'm not sure if there is anything specific to Frigate in this case, or if it's just that Frigate is the most read/write intensive application I'm running.

 

Here are the things I've tried:

• Rebooting (duh)
• Switching to a new USB drive (by downloading a backup of the old drive, so if any data was corrupt it could have been transferred over)
• Deleting the Frigate container and image, and re-configuring it
• Using different ports (USB2 and 3)
• Legacy Boot and UEFI in BIOS
• Parity check (0 errors)
• Both drives I've used were USB2

 

Some things I've noticed while poking around in the logs/interface/terminal:

• `usb 1-8: reset high-speed USB device number 4 using xhci_hcd` starts showing up a few times in the few minutes leading up to the USB disconnect. Before this, there is over an hour of silence in the log, during which Frigate runs with no issue (Nov 26 16:01:52 if anyone starts poking through the log)
• `usb 1-8: device descriptor read/8, error -110` shows up around the same time.
•  As a sanity check, in the "System devices" page, I don't see any device on port 1-8.
• During the startup sequence in the terminal, I see what looks like a bunch of memory addresses, and something along the lines of "Not correcting this error." I can't scroll up in the terminal afaik, so I can't see the full message, and I haven't found this anywhere in the logs yet.

 

So, there are a few things pointing to data corruption on my drive, likely in a file that was preserved when I backed-up and transferred the drive contents to a new drive, and somewhere that isn't accessed until Frigate happens to need it after a few hours. I'm not sure where to go from here.

 

If anyone has any pointers on where I should keep digging, or has the time to dig around in my diagnostics, I would be incredibly grateful.

tower-diagnostics-20231126-1634.zip

23 minutes ago, sdiber said:

Hi all, I'm new around here, so I appreciate any help and debugging tips with a problem that has been driving me crazy for the past week, ever since moving my Unraid instance over to a new computer.

 

My diagnostics are attached.

 

Since switching to a new computer, my unraid instance has been having issues with the USB flash drive, only when I'm running my instance of Frigate. With Frigate off, the server runs for days with no problem, but after starting Frigate, the server runs for an hour or two, and then shows the `USB Device is corrupt or unavailable` error. I'm not sure if there is anything specific to Frigate in this case, or if it's just that Frigate is the most read/write intensive application I'm running.

 

Here are the things I've tried:

• Rebooting (duh)
• Switching to a new USB drive (by downloading a backup of the old drive, so if any data was corrupt it could have been transferred over)
• Deleting the Frigate container and image, and re-configuring it
• Using different ports (USB2 and 3)
• Legacy Boot and UEFI in BIOS
• Parity check (0 errors)
• Both drives I've used were USB2

 

Some things I've noticed while poking around in the logs/interface/terminal:

• `usb 1-8: reset high-speed USB device number 4 using xhci_hcd` starts showing up a few times in the few minutes leading up to the USB disconnect. Before this, there is over an hour of silence in the log, during which Frigate runs with no issue (Nov 26 16:01:52 if anyone starts poking through the log)
• `usb 1-8: device descriptor read/8, error -110` shows up around the same time.
•  As a sanity check, in the "System devices" page, I don't see any device on port 1-8.
• During the startup sequence in the terminal, I see what looks like a bunch of memory addresses, and something along the lines of "Not correcting this error." I can't scroll up in the terminal afaik, so I can't see the full message, and I haven't found this anywhere in the logs yet.

 

So, there are a few things pointing to data corruption on my drive, likely in a file that was preserved when I backed-up and transferred the drive contents to a new drive, and somewhere that isn't accessed until Frigate happens to need it after a few hours. I'm not sure where to go from here.

 

If anyone has any pointers on where I should keep digging, or has the time to dig around in my diagnostics, I would be incredibly grateful.

tower-diagnostics-20231126-1634.zip 224.23 kB · 0 downloads

Do you map any devices into the docker?

1 hour ago, SimonF said:

Do you map any devices into the docker?

Nothing is mapped directly, but Frigate is running in Privileged mode and uses a Coral TPU. It's on USB port 2-7.

I disconnected the Coral and so far, the server has been up for over 2 hours with no issue, so it seems like the Coral is the most likely culprit. Definitely a problem for me though, since I depend on the Coral for my home security cameras.

 

Any ideas why the coral would be causing other USB devices to disconnect? There are some references to port 2-7 in the log, but none around the time of the crash. I also know the Coral is working up until the disconnects, because it shows up in Frigate, and I get event detection for the full hour-or-so that it runs.

Edited November 27, 20232 yr by sdiber

Did you find the solution to this one? I have 2 PCs exactly the same (prod / developement) and (funny enough) the production machine comes up with flash drive error. I'm thinking about swapping the Coral device and see if it makes a difference.

 

Nick