Deployment Completed With Errors

[paul.r.r]

I am working on a project with an sbRIO-9628 and have been fighting with the 'deployment completed with errors' issues for far too long, and need some help or direction from NI.I have created a project that includes a small subset of the code in my real project which can replicate this error - please see attached. When I attempt to run the vi RT Main.vi, I get a variety of error messages, all some flavor of deployment completed with errors - vi loaded with errors on the target or a message that one of my vis is broken. When I navigate to the 'broken' VI, the run arrow (sometimes) does show as broken, but the VI isn't actually broken - if I click the run arrow the error list that pops up shows nothing is broken. Mass compiling fixes the broken run arrow, but the next attempt at deployment usually results in the same problem again - deployment completed with errors, or vi loaded with errors on the target. (Usually a different VI.)

I am using lots of classes and the actor framework, and one VI that I have noticed that is regularly reported as being the vi with errors is an override of receive message, which simply logs the incoming message and then calls the parent method. I've removed my logging call, and kept the override that simply calls the parent method, but I still get the error. If I attempt to run this VI directly, LabVIEW usually crashes. (I have included a crash report from this exact scenario in the crash report folder)

I have also noticed that closing the project doesn't actually clear the project items out of memory - I can close the project, then reopen and the project opens instantly (it usually takes 20+ seconds to load everything on first load) and will show as already being connected to the target. I assume this is somehow related to the problem, but may not be.

I have gotten this code to deploy successfully in the past - it seems as though once it is correctly deployed it works until the code is opened under a windows target, at which point I am back at square one. In order to get it to deploy previously I had to spend about a day mass compiling, clearing compiled object cache, restarting LabVIEW and my PC over and over again attempting to deploy one tiny piece at a time until I got it successfully working. 

I am looking for some direction from NI here - even if they don't have the solution, is there somewhere I can go to get more information about what is actually happening? Can I somehow see what code is and isn't deployed on the target? Can I get a more detailed log of what is happening during deployment that what I see in the deployment window? 

[pedrotorres]

Once I started forcing my VI's to recompile (<Ctrl> + <Shift> + Run Button) and saving all before running I stopped receiving this "Deployment completed with errors" message. This may help you:

https://knowledge.ni.com/KnowledgeArticleDetails?id=kA00Z000000P82uSAC&l=pt-BR

[paul.r.r]

Thanks for the feedback, unfortunately I have tried to mass compile then deploy, clear compiled object cache then deploy, clear compiled object cache, mass compile, then deploy, individually compile vis (including an intelligent compile - dynamically sorting all vis by dependency so I compile the vi with no dependencies first, then compile vis that depend on that one and so on) then deploy with no success. 

I have found that I can often get a successful deployment if I create a blank top level vi, then put the lowest level class on it (class with no dependencies), deploy, then add the next level class (class that depends on the first class) deploy and so on until I build up the main vi that includes everything needed. I am going to write a tool to automate this, but I would really like some help or support from NI on this - if nothing else, give me the vi server methods that I can call to get more details about what went wrong when I try to deploy. See the thread below - at this point, this issue has been a recurring problem for 10 years. What progress has been made? The response several years ago was that this is difficult to debug because its a generic error. Was logging added to every step of this process so that we can get more details about the error? 

https://forums.ni.com/t5/Real-Time-Measurement-and/vi-loaded-with-errors-on-the-target-and-was-close...

[MarkCG]

This kind of problem has been around for a decade but has never been resolved. Engineer ambitiously.

[paul.r.r]

Well, I have completely reformatted my sbRIO and reinstalled all software, and the problem persists. In hopes that could build my own automated solution, I created a tool that will sort my project libraries and classes in order of dependencies. It then adds these classes/libraries to an empty project one by one, deploying after each new module is added. This works fine with no errors. Finally, once all modules have been added to the project, I add my main vi to the project, recompile it, then attempt to deploy. And, I get the exact same error.

In the thread below, someone mentioned this may be related to circular dependencies. (I think it is because the NI deployment process deploys code not in order of dependencies) My code doesn't have circular dependencies, but the actor framework and af debug is a circular dependency, so I removed all of the AF debug code from my copy of the actor framework, with no changes to behavior.

https://forums.ni.com/t5/Real-Time-Measurement-and/vi-loaded-with-errors-on-the-target-and-was-close...

Attached is my MAX log, but it doesn't seem to have much useful information.

[paul.r.r]

Another piece of information that may or may not be relevant/useful. After an unsuccessful deployment attempt, I have noticed all my class icons in the class hierarchy window have little pins (and one file) icon overlaid on them. See attached - the little grey marks on the icons are not mine.

[ChrisStrykesAgain]

Is it possible to replicate the crash using the .zip you provided in the first post?

[paul.r.r]

---

@ChrisStrykesAgain wrote:

Is it possible to replicate the crash using the .zip you provided in the first post?

---

Yep, the code provided in the first post causes this crash fairly consistently on two independent sbRIOs (9628s). If you have an RT device and can try it out, that would be greatly appreciated.

I should clarify - I generally get the deployment completed with errors message the first try, and if I try again immediately after, labview usually crashes.

[HB@BNT]

Did you try to clear compiled object cache?

That was the solution for me in one case in the past.

Maybe it's a solution for you, too.

Best Holger

[Hooovahh]

I might have missed it but what version of LabVIEW are you using?  I saw lots of deployment errors in 2017, getting better in 2018, and now I haven't seen those types of errors in 2020 yet.  When I did see it I would delete all the compile cache, re open and recompile everything and then it would work for a bit.  In 2017 I remember that the issue persisted to builds too.

I would make a build, not ever deploying it.  It would make the RT EXE, I would then copy it over and restart the controller but the RT EXE was broken in some way and nothing would run.  I'd delete the compile cache, re open, rebuild, copy the new RT EXE over and it would run just fine.  This was concerning because it made an RT EXE successfully and it thought it was good but it wasn't.