This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.
> Full Name: Manh Ha NGUYEN > Organization: Laboratoire des Sciences du Climate et de l'Environnement, > France > Package Version: NetCdf 4.3.2 with hdf5 1.8.11 parallel > Operating System: Fedora 20 > Hardware: Intel Xeon > Description of problem: Hi, > I am working with NetCdf 4.3.2 with hdf5 1.8.11. > My process creates netcdf file, receives data in each time step and writes > them to the file. > This program is made to run for several days. > This program has worked well with a small number of time steps. > However, there is one problem when I try to make the program run longer: The > memory increases and the program explodes. > If I don't use nc_put_vara_xxx functions (just comment them in my code), > there is no "memory leak" any more. > On using valgrind, with nc_put_vara functions activated and unactivated, I > see no memory leak. > I have impression that nc_put_var_xxx functions use a temporary memory which > increases each or after several writes and is only released on closing netcdf > file. > > Could you give me a suggestion how to get around this problem? > > P.S: In fact, there is a thread about this phenomenon of netcdf [netCDF > #XIX-708449] Yes, but that thread showed that the memory leak fix in version 4.1.2-rc1 fixed the problem for that user, so that there were no longer any memory leaks after the memory use reached 23 MB (or 21 MB if he nc_sync() was called). Since you are using version 4.3.2, that fix is in your library, so while your problem may have similar symptoms, I think it is a different problem. Can you provide us with a complete example that demonstrates the memory leak? That would really help to debug the problem. If we can't duplicate it here, debugging such problems from just a description is very difficult. --Russ Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu Ticket Details =================== Ticket ID: XIX-708449 Department: Support netCDF Priority: Normal Status: Closed