BadEncodingLimitsExceeded #434
Comments
|
what do you mean by publish? Do you mean a subscription? |
|
By "publish" I mean adding variables to my server, via |
|
Is the browse request failing? Did you try to tweak the buffers? |
|
First I get a “Browse succeeded”, but inmediatly after I get “Browse failed”. Here it is the whose sequence of messages displayed by the client: 10:33:15.808 | Server Node | open62541 - None - None (ua... | Connection status of server 'open62541 - None - None (uatcp-uasc-uabinary)' changed to 'Connected'. The server also gets a message: I understand that is because REPUBLISHREQUEST is not yet implemented (issue 370). Am I right? On the other hand, about tweaking the buffers, I’ve tried this: Making buffers 100 times larger, but I get the same result. De: Sten Grüner [mailto:notifications@github.com] Is the browse request failing? Did you try to tweak the buffers? — |
|
Yea the second error |
|
Thanks. Now I also changed the line you suggested: But I get the same error. Any idea? |
|
can you use wireshark to see if the response is actually coming? It can be that it is bigger than UaExpert can accept |
|
Hi Sten. With fewer variables (80) everything works ok. De: Sten Grüner [mailto:notifications@github.com] can you use wireshark to see if the response is actually coming? — |
|
Well, hook up wireshark https://github.com/acplt/open62541/wiki/Capturing-traffic-with-Wireshark to see if the browse response is actually coming from the server. My theory is that we try to send something that is bigger than UaExpert announced it can accept. The problem is the missing chunking mechanism. |
|
Hello. I've used wireshark, and the browse response is actually coming from the server. Anyway, my problem is partially solved. Why? Before all the variables where published at the same level (all of them in a single flat level). Anyway the problema is still there. If your ask for too many variables in a "level", UaExpert asks for a Republish, and as this is not implemented yet in 62541, It is not able to browse the variables. |
|
No, I think you misunderstood it, the problem is the _[Malformed Packet]_, having no publish request is okay, since the standard does not force you to implement it. |
|
@osaiz i can not reproduce your error :( i have added 2000 nodes, and it is browsing, which uaexpert are you using? How can I reproduce the error, can I have the wireshark trace and code of the server? |
|
I’m using version 1.02.334.206 for Windows. I get also a problem with another OPC-UA client: Prosys-OPC-UA-Client. But I may be wrong… De: Sten Grüner [mailto:notifications@github.com] @osaizhttps://github.com/osaiz i can not reproduce your error :( i have added 200000 nodes, and it is browsing, which uaexpert are you using? — |
|
Well, with 1.3.1 I cannot reproduce it anymore, but the problem is on our side... i will give it a try in Prosys-OPC-UA if i find it |
|
Sorry Sten. I’ll try to make a simple example reproducing the error and sending it to you De: Sten Grüner [mailto:notifications@github.com] Well, with 1.3.1 I cannot reproduce it anymore, but the problem is on our side... i will give it a try in Prosys-OPC-UA if i find it — |
|
Please find the example that reproduces the error. If you delete one call (just make 95 calls), it works ok. De: Sten Grüner [mailto:notifications@github.com] Well, with 1.3.1 I cannot reproduce it anymore, but the problem is on our side... i will give it a try in Prosys-OPC-UA if i find it — |
|
where can i find the code/binary? you can just paste the code into pastebin |
|
I cannot attach .c files. De: Sten Grüner [mailto:notifications@github.com] where can i find the code/binary? you can just paste the code into pastebin — |
|
@Stasik0 meant something like this: http://pastebin.com/ |
|
Thank you. De: FlorianPalm [mailto:notifications@github.com] @Stasik0https://github.com/Stasik0 meant something like this: http://pastebin.com/ — |
|
Thank you. |
|
@osaiz i cannot compile your code with the current head, can you tell me which version are you using? (it is in the 3rd line of open62541.h) |
|
Hey @osaiz Works fine on my box. Was that the version with the error? |
|
|
@osaiz i cannot reproduce it either, I have just modified your code slightly, are you testing over via localhost? Or do you have some networking between the server and the client? Here is the code i test: http://paste2.org/MbDJh1nD, with v0.1.0-RC4-614-g32c5050 |
|
I'm not testing on localhost. My server is in a Linux machine and the client on a Windows machine. I am using the amalgamated version. This is my compilng command: |
|
The problem seems to be around "browseName" parameter of UA_Server_addDataSourceVariableNode. With shorter values of strings for this parameter everything works ok. |
|
is the code you have posted working or not? If I run it on linux i can browse from windows over ethernet using both softing and uaexpert clients you can see the screenshot from wireshark, i get the 97 responses of browse in about 13000 bytes |
|
No, it is not. :-( Alternativelly, if I create one variable less, or if the browseNames of the variables are shorter, then it does work correctly. |
|
you run the same code I posted (http://paste2.org/MbDJh1nD)? Could you please attach a wireshark log? The capture file with the error? |
|
Yes. I'm usign your code. I've just captured the whireshark log. |
|
something like http://www.tinyupload.com/? |
|
@osaiz okay, if you look into the packet 131, you will see the 97 references that are coming safely, but then, at the end of the message there is a DiagnosticInfo array, which size is set to "ArraySize: 256002305" that is definitely invalid and breaks everything. Since we cannot reproduce it, I'd say it can be a nonititalized memory allocation or somehting similar, can you reproduce the error with the current head? and code of @jpfr http://paste2.org/yKGUbbgy? |
|
@osaiz @jpfr there are more problems, look into these screenshots The first on is from @oasiz capture.. there the name of the last item is being mixed up with "MSGF", which seems to be a new or old message? How does it get in there?
The second screenshot is how it has to look like -> my machine It seems that the response is somehow overwritten by the next message (wut?)? |
|
Hello. I did it yesterday (versión 0.1.0RC4-714--g106bf77). Is this the last one? Or should I check with another one? By the way, this version does not compile with -DENABLE_SUBSCRIPTIONS. It gives compilation errors.
Here you are the wireshark logs of both executions. http://wikisend.com/download/780370/right.pcapng Right execution About the last post by @Stasik0, I have no clue about how that gets corrupted, but please notice that it happens in the last variable (the one that makes things fail). |
|
@osaiz yes, this is already the newer encoding code... the problem is that I can not reproduce it at all :( which compiler and kernel verision are you using? Could you please try to reproduce it via localhost (there is UaExpert for linux too). |
|
gcc 4.4.1 I'll try to reproduce it in a more modern Linux, and with UaExpert in local |
|
I've tested in a newer Linux, and the problem disappeared gcc 4.8.3 |
|
This is still very creepy....
|
|
okay, i will close it |
|
This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs. |
I am writing a OPC-UA server based on open62541.
If I publish a limited number of variables (let's say 80), it goes OK.
But, if I try to publish a larger number of variables (let's say 200) it does not.
When I connect my server with UaExpert, the variables are not displayed, as I get "BadEncodingLimitsExceeded" error.
What could I do?
Pd.- I'm using trunk version of 2015/09/28
The text was updated successfully, but these errors were encountered: