when you will fixes Unknown error occurred?

[10der 832]

Please login or register to see this image.

/monthly_2022_01/image.png.86402ff31cb247ba4823c7607ccc2eff.png" />

 

 

is this too hard?

[jgab 2,642]

Is it a http request issue again? Is it possible to create a MVP that causes the error? I don't think I have seen this at all...

[10der 832]

Jan, Fibaro has developers Fibaro has QA let’s chance to Fibaro to fix this bug 

I am very tired… Now I have only 5 zwave light switches and 5 aeot 6 multi sensors what I can’t replace easily with zigbee 

I am do not care about HC3 tbh at all and do not want spend time on MVP

sorry 

[petergebruers 1,268]

I contacted support because one of my long running QA had this issue after 11 days uptime of last beta. It is a QA that rarely shows this issue, I have been using this for more than a year. It is indeed a http QA in a way because it uses "refreshStates". Not your average thing. Support replied that they have paid support for that and that's where it ends for me. I thought they would be pleased with having a logsdump + opportunity to do a live investigation of the problem. Not really............

Edited January 7, 2022 by petergebruers

[SmartHomeEddy 1,005]

1 minute ago, petergebruers said:

Support replied that they have paid support for that

???

[petergebruers 1,268]

4 minutes ago, SmartHomeEddy said:

???

Yeah well they saw "quick app" and I understand... first line support has to act fast and make decisions.

 

As someone pointed out in private... "Helpdesk does not have means nor time to launch gdb (gnu debugger) on your HC to investigate"

 

And I didn't have the patience to "nudge" someone else at Fibaro. I rebooted my HC, that always fixes the issue...

 

Edited January 7, 2022 by petergebruers

[SmartHomeEddy 1,005]

They should pay you 

[10der 832]

2 hours ago, petergebruers said:

Support replied that they have paid support for that

WUT? ? 

FuckIBARO the best ever.

2 hours ago, petergebruers said:

Helpdesk does not have means nor time to launch gdb (gnu debugger) on your HC to investigate"

lets ask Tinman

he knows gdb as old man like you or me ?

but one issue on my side for example i am stoped hacking (read as debugging) on 32 bit asm in SoftIce ? 

the next step in my life was c++ ? 

[tinman 1,070]

1 hour ago, SmartHomeEddy said:

They should pay you 

 

when you buy car, it is clear that you have to pay each time you ask to fix issue in that car. Yes you will get always some kind of "support", but at the end of the day, everybody knows that service cost money. It is complex machine, and even very obvious things are paid by us, not the car manufacturer. Smart home systems are not toys, they are very complex by default, and any addition we add to them makes things much more unrpredictable and hard to track without being on that specific system, sometimes for hours. 

 

 

8 minutes ago, 10der said:

FuckIBARO the best ever.

 

Oleg, be careful, Putin is watching you ?

[dlizak 7]

4 godziny temu, tinman napisał:

when you buy car, it is clear that you have to pay each time you ask to fix issue in that car

What? What about warranty, recalls? If there is safety issue, manufacturers sometime fixes these even after warranty period. The fact is, HC3 was not ready for production on launch. So if you buy a product that is broken straight away, it's pretty obvious you want it fixed.

[jgab 2,642]

It's a bit tricky.

Is it like with the "

Please login or register to see this link.

", that it's likely a developer bug but it's just badly reported by the framework? I still think they should fix the error reporting - because it's an "error in the developer experience" - but they don't seem to prioritise that aspect.

 

If it's an error that only appears for some people once in a blue moon it adds to the difficulty. If my program runs for 3 years and then the bug appears still means it can be my bug... (it usually is)

As a developer, it's usually up to me to prove that my code is ok before I can blame the framework - unless I want to pay them for debugging my code - because if it turns out to be like the "empty error" and caused by a user bug but badly reported - who to blame?

 

That's why I asked if there was a "minimal viable program" that exhibited the "unknown error"...

 

Now I found that @10der's brother @hater did post a

Please login or register to see this link.

that caused the error for him - unfortunately not for others. Then the scope for the bug hunt is wider then just the code - how the system is configured - what else is running -  how the QA has been updated etc etc.

Now it also seems like @m.roszak thought

Please login or register to see this link.

giving the cryptic clue:

Quote

Hah, we already know what is happening - we need good solution to report this to the UI 
So - maybe we will change the log...

which suggests that they could probably produce a more specific error message (knowing what subsystem caused it) but they didn't have the solution to get the exact info to log (I'm guessing now). It also gives a sense that it's caused by some "developer procedure" that they can't prevent so they need to report it. If it was a framework bug they should just fix it and no need to log?

But yes, I believe that they could share what they believe cause the error so that we as developers can try to avoid it - while we keep our breath waiting for better error messages.

 

 

 

Edited January 8, 2022 by jgab

[10der 832]

9 hours ago, tinman said:

Oleg, be careful, Putin is watching you ?

 

[petergebruers 1,268]

5 hours ago, jgab said:

That's why I asked if there was a "minimal viable program" that exhibited the "unknown error"...

 

Well, that was kind of the whole point why I reported it to helpdesk, because it is hard to reproduce and my HC3 had such QA in such a "weird state". It happens 3 times per year. I offered Fibaro a way to look into this and they said "you have to use paid support for that".

 

So maybe 3 times per year on my system: a QA starts "restarting" itself and keeps saying "Unknown error occurred:"

 

If it happens, which is rare, one QA stores data in an InfluxDB and the other one uses "refreshStates". When it happens to one of this WAs AFAIK then no other scene or QA suffer from this.

 

Sometimes it is possible to just "SAVE" a QA (maybe make some minimal change) and it will be OK. But frankly, the fastest way to get everything going is just reboot the HC.

 

12 hours ago, tinman said:

we add to them makes things much more unrpredictable and hard to track without being on that specific system, sometimes for hours. 

 

Yup and that is why I did not see my mail as "asking for help" but as an opportunity for Fibaro to improve their system. That did not work.

 

Maybe I should have been clearer about my intentions...

 

Please login or register to see this code.

 

That wasn't pressing anybody, right? Did I say something wrong?

 

Here is the full, unedited answer I got from support:

 

Please login or register to see this code.

 

I honestly did not know what "more@fibaro..." was so I had to ask but after several days I got no answer back

 

Please login or register to see this code.

 

Not sure why I never got an answer to that, do you guys think I said something wrong? 

 

I had to ask my friends here...

 

Anyway... I really do believe it is a weird bug and very likely a rare one because only a handful of users report it on the forum, so yeah, I all I want to say is

 

II tried to cooperate with Fibaro on this and that did not work this time. We probably misunderstood what this is about...

 

When we have a "programmable" system the boundaries always vague. I understand nobodies time is "for free" and we all have some responsibilities.

 

Imagine this was an open source system: would I be able to get in touch with someone? Would I be able to run a "debugger" myself? Would I be able to find the core issue, either in my code or in some system code? And then again, would that knowledge lead me to fix the issue or get someone motivated to fix it for me?

 

The answers are not at all clear to me. What is clear to me, regarding the problem reported by @10der and me (the "Unknown error occurred") we are not cooperating to solve it.

 

But I don't care enough to change things in this respect.

[10der 832]

Please login or register to see this code.

in Ukrainian and  Polish as well we have word "mor(z)e". if translate in english....

a Fibaro support hint you rest on the sea  

 

[SmartHomeEddy 1,005]

41 minutes ago, petergebruers said:

Not sure why I never got an answer to that, do you guys think I said something wrong? 

No, nothing wrong with that. I also thought Fibaro would be helped with your findings. But somehow Fibaro reacts (or does not react at all) in a way I don’t understand ?

[jgab 2,642]

The

Please login or register to see this code.

errors we usually get when we call some C/C++ function with wrong arguments.

Ex.

Please login or register to see this code.

Please login or register to see this code.

but we usually get a luabind error message like above. So there seems to be some function (builtin) that we or Fibaro calls that sometimes or in some circumstances doesn't provide the error message - or they accidentally drop the error message.

[10der 832]

btw Jan

after this error at 90% - got 502 and goto rebooting HC3

 

i think it`s a memory leak issues.

Edited January 8, 2022 by 10der

[petergebruers 1,268]

10 minutes ago, jgab said:

So there seems to be some function (builtin) that we or Fibaro calls that sometimes or in some circumstances doesn't provide the error message - or they accidentally drop the error message.

Yes. The interesting is, imho, the fact that the "restarting mechanism" of the QA keeps hitting the same problem (and same error message)... Until you reboot. So either it is not a Linux Process that gets restarted (because imho the Linux kernel is good at cleaning up a process)... Or something is shared (and exhausted) between processes. Or the API server is out of "something".

 

I think it is difficult to understand without knowing the pieces of the puzzle...

 

I have been speculating... Both mqtt and http QAs seem vulnerable.

 

Surely CPP exceptions and luabind are possible (confounding) factors

 

I don't even know "who or what" is supervising that QA and reporting the unknown error.

[tinman 1,070]

27 minutes ago, jgab said:

The

Please login or register to see this code.

errors we usually get when we call some C/C++ function with wrong arguments.

Ex.

 

as far i understood from IDA session, the lua engine get lua code piped from quickapps, when the code is incomplete or corrupt, the unknows error occurs. So the question is more, is the quickapp code broke because something caused overflow and data corruption, or the lua code itself contains characters which breaks the pipe

 

Please login or register to see this image.

/monthly_2022_01/image.png.d47a9fa17a6f4e81d71cc7e440a0fa1f.png" />

Edited January 8, 2022 by tinman

[10der 832]

1 minute ago, tinman said:

 

as far i understood from IDA session, the lua engine get lua code piped from quickapps, when the code is incomplete or corrupt, the unknows error occurs. So the question is more, is the quickapp code brokek because something caused overflow and data corruption, or the code contains characters which breaks the pipe

not sure. it`s look like yes and no.

IMHO issue between code HC3 and LUAbind like a memory leak 

 

btw, this error was in my sonos hub. 

so, yes HTTP calls in QA