TOPIC:

I don't know if anyone on this thread is still paying attention, but I think this memory issue is fixed in version 1.2.1.2, currently in beta.

Jason

I'm having the same problem. I crash after logging data for a few days. Has this ever been fixed?

H5Dwrite (Variant).vi<ERR>
An HDF5 Error Occured
Major: Dataset, Minor: Write failed.
can't prepare for writing data
(Major: C000000, Minor: C000021)

HDF5 Error Stack: {{C000000;C000021;H5Dwrite;..\..\src\H5Dio.c;225;can't prepare for writing data},{C000000;C000021;H5D__pre_write;..\..\src\H5Dio.c;347;can't write data},{C000000;C000035;H5D__write;..\..\src\H5Dio.c;639;unable to set up type info},{C00000C;C000025;H5D__typeinfo_init;..\..\src\H5Dio.c;1040;memory allocation failed for background conversion},{C00000C;C000025;H5FL_blk_calloc;..\..\src\H5FL.c;954;memory allocation failed},{C00000C;C000025;H5FL_blk_malloc;..\..\src\H5FL.c;886;memory allocation failed for chunk},{C00000C;C000025;H5FL_malloc;..\..\src\H5FL.c;206;memory allocation failed for chunk},}


<b>Complete call chain:</b>
H5ErrorHandler.vi
H5Dwrite (Variant).vi
Append Element(s) to Dataset (Variant).vi
Save data to File.vi
Save Data HDF5 single.vi
temp.vi
Vibe system 5.0 single.vi

Hi Andre,

I apologize, but I have not tracked it down. Given my current schedule, and the tenacity of this bug, I'm having a hard time finding enough time to dedicate to this problem. I'll keep you posted.

Jason

Dear Jason, I wanted to ask you if you already have found the bug in the LVH5DwriteVariant function?

Hi Andres,

Just an update. First of all, I've reproduced the problem and I've narrowed it down to the call to LVH5DwriteVariant which is in my helper library. This is good and bad. Good, because there is actually a lot of other things that happen when the high level functions are called that I was worried were the source of the problem. Bad because I now need to go through that non-insubstantial function with a fine-tooth comb and see if I can't narrow down the leak. It is still possible that the leak is in the underlying HDF5 library as well, though, frankly, that seems far less likely.

Hopefully I will learn more this week and can nail down the source of the problem.

Jason

Dear Jason

option 1) caused my program to fail. After running H5close (and H5open) I had to completely restart LabVIEW

option 2) didn't help either. I didn't play that much with the chunk size, but the memory increase was not affected in a positive way. With my own data cluster, which consists of about 33'000 values, the memory increases by approx. 300KB per storage procedure or 2.6GB per day (with a sample period of once every 10 seconds). That is even more than what my file grows in this time.

I produce daily files, hoping, the memory would be released after storing into a new file. But this is not the case.
Have a nice weekend.
Best Regards,
Andres