Data created when registering MSDK 5.8.0 blocks future DJI Dronelink launches
Completed
Written on April 18, 2024 11:30 UTC
Since February 2024, the Dronelink DJI 4.9.0 application closes when the camera's video stream is displayed.
Dronelink DJI has been using DJI MSDK version 5.8.0 since December 13, 2023. This Dronelink DJI version 4.8.3 (278) did not have this bug.
Since mid-March, the bug seems to have changed for the better. The table below summarizes the various tests carried out on the Dronelink DJI 4.9.1 (290) application. Version 4.9.2 (291), which is the most recent on April 18, 2024, has the same bug.
A screen recording can be downloaded from Youtube to illustrate six tests: https://youtu.be/U4djOeQZ6qI
At the end of tests 1, 4 and 6, it became possible to display the video stream by acting on the data just before each application launch. Unfortunately, this means that Dronelink DJI cannot be launched in the absence of a network (tests 3 and 5). What's more, before each launch, you have to clear the data created during the previous session.
The right-hand column shows the memory sizes used by the data. These sizes are recorded when Dronelink DJI is closed after each test.
This bug is experienced by other Dronelink users, mainly in France. The difficulty for Dronelink is that no user in the USA has encountered this bug. Of course, a discussion thread has also been created on the Dronelink forum
Here's Dronelink's most recent response for this topic:
"We have not changed any code between 278 and 279 that would have affected the data in this folder, and all the data in the folder is generated by the DJI SDK.
The logic that it's not the DJI SDK is only partially valid - the problem is that DJI SDK registration uses a network server hosted by DJI, and the payload of data coming back from that server can and has changed over time due to changes DJI has made to the code running on the server.
Without being able to reproduce the problem, it's actually very difficult for us to identify where the bug lies (whether in our code or theirs)..."
Which leads to the following questions:
How can DJI make the software used on the DJI servers in charge of registering the SDK in France be "lightened" so that the camera's video stream display works the next time DJI Dronelink is launched?
What elements would DJI need to identify the origin of this bug?
Hardware and software used for these tests :
DJI mini 3 drone; firmware: v01.00.0410
RC-N1 radio control; firmware v04.16.0500
DroneLink DJI Version 4.9.2 (291) ; Kernel 4.9.2
-
If you encounter a crash, could you provide us with the crash logs? Based on your test results, it appears that Dronelink can successfully retrieve the drone's video feed after clearing the cache. However, the sample code from MSDK (MSDK's standard testing tool) can still retrieve the video feed without clearing the cache. Therefore, for the issue you are experiencing, it would be very helpful if you could provide us with a crash log. When you reproduce the crash on Dronelink, if your device has developer options enabled and your computer has debugging tools such as ADB installed, you can use the command "adb logcat -b crash" in the command window to print out the crash logs that just occurred. The crash logs are crucial information for us to identify the cause of the crash. Version 5.9 is expected to be released in April or May. -
Thank you for your prompt response.
Eight logcat incident files have been created and placed in a compressed folder which can be downloaded at: https://transfert.free.fr/zhmiF5W
This link is valid until May 20, 2024.
The files concern each of the 6 tests mentioned in the topic.
Test 1: logcat_04-21-2024_09-27-46test01ok.txt
The version of Dronelink DJI downloaded on April 18, 2024 is number 4.9.2 (291) | Kernel 4.9.3
Test 2: logcat_04-21-2024_09-29-49test02crash.txt
This test ended with a crash.Test 3: logcat_04-21-2024_09-33-01test03crash.txt
This test also ended whith a crash. Android suggested clearing the cache.
A new test was performed and also ended in a crash.
It was found that clearing the cache did not change the result.Test 5: logcat_04-21-2024_09-35-19test05dskToRegister.txt
The application was closed after displaying the request to connect to the network to register the DJI SDK.Test 6: logcat_04-21-2024_09-37-02test06ok.txt
The test ended successfully.Test 4:
logcat_04-21-2024_09-40-30locPreciseTest04ok.txt et
logcat_04-21-2024_09-42-36locPreciseTest04ok.txt
These two tests were run with a precise location selection.
Both of these tests ended correctly, while previously precise localization led to a crash.Please feel free to provide further information to help get DJI Dronelink working for a DJI mini 3.
Hardware and software used for these tests
DJI mini 3 drone; firmware: v01.00.0410
RC-N1 remote control; firmware v04.16.0500
DroneLink DJI Version 4.9.2 (291) ; Kernel 4.9.3
Android smartphone :
Samsung Galaxy A34 5G sans carte SIM ;
wifi : Free, Orange ou SFR
code du modèle : SM-A346BZKAEUC
Android 14 ; OneUI 6.0 ;
Build : UP1A.231005.007.A346BXXSBWL2 -
I am unable to access the link you provided. Could you please upload these files to this address? https://pan-sec.djicorp.com/s/nGJi54WS2X2ZgiZ password:dji123 -
Here's an additional video detailing the various steps of these 6 tests: https://youtu.be/bjHasTPaRIY
-
I have received the logs you provided me. However, these logs do not contain any crash records, or even SDK and DroneLink log information. Have you tried capturing the crashes using the methods I mentioned above？ You can also install the crash capture tool to help you capture crash information. -
April 25, 2024 09:30 UTC
Here are two logcat files obtained with Android Studio linked to the Samsung A34 during tests 2: without deleting data before launching Dronelink DJI.
They have been uploaded in the space reserved for this purpose:
https://pan-link.djicorp.com/s/nGJi54WS2X2ZgiZ
As using Android Studio is a discovery, please let me know if these logcat files are sufficient to get an idea of the origin of the bug.
-
April 25, 2024 15:45 UTC
Here's another logcat file generated with Android Studio that shows an installation test (test 1) that displays the camera's video stream correctly.
This test is followed by 3 attempts at test 2, which consists of relaunching the Dronelink DJI application without first deleting the data. Test 2 tests all lead to the same bug.
This file is entitled: samsung-SM-A346B-Android-14_2024-04-25_173217.logcat
-
Greetings! I have found the storage path of the crash files in your documents and need you to provide us with these crash files. I have noticed a similar issue in the system with ticket number 105405. Can we reply to that ticket? Attachments cannot be added in the forum, but they can be added in the ticket. -
1er mai 2024 13 :45 UTC
Bonjour,
Merci pour vos suggestions transmises à Dronelink. Dronelink a proposé une version beta qui fonctionne correctement pour les tests 2 et 3 qui sont maintenant sans bug.
Concernant ce bug, il restera quelques mystères :
- pourquoi le bug est apparu exactement le même jour (3 février en France) que lors de l’upgrade vers la version 4.9.0 de Dronelink DJI ?
- pourquoi le bug ne s’est manifesté que sur le sol français ?
En tout cas, la résolution de ce bug est la bienvenue car Dronelink saura, maintenant, comment le résoudre pour les versions ultérieures de son application, même s’il ne se manifeste pas aux USA.
Cordialement,
-
- pourquoi le bug est apparu exactement le même jour (3 février en France) que lors de l’upgrade vers la version 4.9.0 de Dronelink DJI ? --> I am unsure of the differences between the two versions of the DroneLink App. I apologize for not being able to answer your question directly. However, I can explain that the workaround we provide for DroneLink actually bypasses a crash in the registration of the MSDK. Currently, we have only received feedback from users in the European region regarding this issue. It is for this reason that we have confirmed the problem you are experiencing after receiving the crash logs. - pourquoi le bug ne s’est manifesté que sur le sol français ? -->This is what we are going to analyze next, and I have not received the exact analysis results yet.
Please sign in to leave a comment.
Comments
11 comments