The valgrind error was not related after all. We found out that we didn't free a memory context where we should, which resulted in memory growth if the application requesting netgroups was a long-running one.
1) Run a process that requests a netgroup periodically but doesn't exit between the attempts. Maybe a deamon, python interpreter might do as well.
2) Watch the sssd_nss memory footprint grow
Created attachment 832692 [details]
source code of reproducer
Bug can be reproduced with attached example program.
gcc -Wall netgroup.c -o netgroup
Memory leak is reproducible also with non-existing netgroup.
RSS of sssd_nss process was increased about 200 MB in 3 minutes.
* master: 65f4003b93157b32eb0dcd6955d37dd009dc960e
* sssd-1-11: 8696135e659685394937f81136521c676f0a0678
* sssd-1-9: 34728310b600c700188b2679118a1af85bcdaaaf
The bug has been fixed upstream in all supported branches.
Tested with sssd-1.11.6-12.el6.x86_64
Verified via beaker run.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.