Device service throws "failed to find device <deviceName> in cache" ERROR

[sounak-1989]

Hi guys,

I am using Device Service SDK to create a new service for my SPI based devices (basically Input and output devices in SPI daisy chain). I am using the "Auto Discover" feature of the SDK.

After successful discovery my autoDiscover() returns the deviceModel structure which is then passed to the deviceCh channel which is of type []sdkModels.DiscoveredDevice.

I am following the Device LLRP as a reference for this design. But Whenever I run my service it shows that the devices are successfully added to metadata and I can query metadata to see the devices. But on the console logs it throws an error "failed to find device in cache". The source of this error is the manageddevice.go file.

Also when I am trying to send a command to my device I get the JSON response "device not found" although it shows the device when I query metadata. Not sure how this is happening. Since most of the stuff is happening inside the SDK I am not able to put breakpoints from within my device service project.

Any help is much appreciated.

[lenny-goodell]

Check Core Metadata logs for error calling device service callback to Add Device.

This will occur if you are running in hybrid mode (device service from command line and the other EdgeX service in Docker) since your device service config has the Service Host set to localhost. Change the config to the IP of your system or simple stop and rerun the device service which will then find the device in Core Metadata at start-up.

[sounak-1989]

Hi guys,

Figured out what the problem was. In my driver.go file I was trying to print a structure which is nil. Hence it throws the error and panics. This brings me to my initial problem. since none of the newly discovered devices are locally stored I am still getting the panic since my device service is not able to find the device locally. I will investigate further into it and keep you guys posted.

Thanks for all the help.

[cloudxxx8]

please check the log of core-metadata to confirm whether it can callback to the device service succesfully

[sounak-1989]

please check the log of core-metadata to confirm whether it can callback to the device service successfully

Yes core-metadata can callback to the device. I can confirm this because when I try to delete the device through core metadata I can see the delete request coming to the device service and the device getting deleted.
Although I have a question

• Is the device service supposed to create new local device files when auto discovery is turned on?

[cloudxxx8]

no, the container should be read-only, so there is no local files change.

in this case, why do you say the cache is not updated? the device service cache are updated through the callback. how did you confirm it?

[sounak-1989]

no, the container should be read-only, so there is no local files change.

in this case, why do you say the cache is not updated? the device service cache are updated through the callback. how did you confirm it?

I am not running my device service from a container. I am working hybrid. So I was expecting the service to create new device files under cmd/res/devices/

I was initially asking about the cache because my device service was throwing an error which I mentioned on the topic. But now that issue is resolved once it properly started communicating with the other services.

Things are looking good now. The only problems are on my side of the driver code. I can see all the communications happening properly when I put breakpoints through the code.