optimden
optimden
> I do not know exactly where you get the data (it should not be _netopeer2_) > > ``` > match-all-filter > ``` > > but they are wrong if...
I attach valgrind massif output files for both sysrepo-plugind and netopeer2-server. Is it acceptable and enough? It is for first time after both programs started. [massif_sysrepo_plugind.txt](https://github.com/CESNET/libyang/files/13661719/massif_sysrepo_plugind.txt) [massif_netopeer2_server.txt](https://github.com/CESNET/libyang/files/13661720/massif_netopeer2_server.txt)
> What exactly are the binaries doing? Generally, the `possibly lost` `valgrind` memory is due to type sharing in _libyang_ which is hard to track. I'll run again stability tests...
@michalvasko , I've got new heap traces for two cases: 1. Periodically sending get and get-config request for different yang modules during 40 minutes: [massif.out.sysrepo_1.txt](https://github.com/CESNET/libyang/files/13712028/massif.out.sysrepo_1.txt) [massif.out.netopeer_1.txt](https://github.com/CESNET/libyang/files/13712029/massif.out.netopeer_1.txt) 2. Performed two get...
@michalvasko , I have now valgrind report with a few definitely losts. But there are a lot of question marks in stacktrace. It seems it's not easy to fix it....
@michalvasko Hello, sorry for a long delay for this issue. I got valgrind report with a lot of "still reachable" losts in libyang code (not from our project files). Please,...
> It may still be caused by problem in your plugins and most likely is, there are no such problems reported normally. You can try to get more callers in...
@michalvasko , i investigated, that memory leaks problem was due accumulate error messages in libyang logging system. Problem function here is "vasprintf". Due massif showed me stacktrace for this function...
Okey, thank you!
I found, that problem was gone when i rollback to sysrepo version 2.2.33 (a1075ab5a5563f4acc21218ac861b4e8e20457b9). The next commit (d0816444941534009a98) changes direct reading and writing into shm for subscription to ATOMIC operations...