Deployment Failures

[Crossle86]

Testing the latest 2024 beta release (and vendordeps) I am seeing a serious problem with failed deployments. After turning the robot on, the first deployment always succeeds. After that there may be one or two successful deployments and then further deployments complete successfully in as shown in the VSCode progress window and the riolog opens but is blank. On the DS, the robot code led is red. Further deployments all fail. Rebooting the RIO sometimes fixes this and sometimes you have to power cycle. After a reboot or cycle, the code does come up and is the code that was last deployed but failed to start.

Further, every few successful deployments, when the code runs it experiences errors on can bus devices. Redeploy cures this code runs normally. Next deployment we are back to blank riolog and no robot code led red. On continued attempts to deploy is will see some random output from our tracing facility in the riolog but robot code led still red.

Finally thought to do a gradle clean. After the clean, the code consistenly deploys, but the riolog contains left over output from an earlier deploy (the random output mentioned above, its the same output each time) followed by our logging showing successful start up.

It appears that things appear to be ok now (other than garbage at start of riolog), so maybe the clean removed something that was causing the problem.

Update: after repeated successful deploys after clean, I did get one where the code started ok but then got lots of errors from CAN devices. redeployed and code ran fine. Deployed again and CAN errors. Deployed again, no errors. After almost 50 deployments trying to nail this down I am calling it a day.

Code is at https://github.com/ORF-4450/Robot24B

[GrahamSH-LLK]

I've been getting this too. It happened a lot less when I disabled robot code and let it sit for a second before redeploying.

[PeterJohnson]

Probably a duplicate of #39?

[Crossle86]

May well be but we don't get that runtime error. I will try the ssh solution and see what we get.

[srimanachanta]

Probably a duplicate of #39?

We are getting the same issue on the roborio2.0 but we aren't seeing any prints regarding memory in DS. I haven't traced what fixes the issue but we usually restart the Rio and then redeploy. I can try SSHing into the rio and running the program like that and see if it is isolated to the deployment system or the running of the robot program.

[sciencewhiz]

Does this still occur with Beta 4?