ROS Service/Client Memory leak?

edit

I've observed this memory leak with roscpp services as well. I noiced that the memory usage of a node which was making service calls at 20 Hz rose from 50 MB to over 10 GB when I let my code run continuously over the weekend. I created a simplified test case consisting of two nodes, one which just advertises a service and one that calls that service at 20 Hz. I ran them under valgrind for 5 minutes. I can't upload a tgz of my package, so I've added my test nodes and the valgrind results here. I've also made an issue on the roscpp issue tracker.

I am using Ubuntu 14.04 with ROS Indigo and valgrind version valgrind-3.10.0.SVN.

This is the client code:

// leaky_client.cpp    
#include <ros/ros.h>
#include <std_srvs/Empty.h>
int main(int argc, char** argv) {
  ros::init(argc, argv, "LeakyClient");
  ros::NodeHandle nh;
  ros::ServiceClient serviceClient = nh.serviceClient<std_srvs::Empty>("/EmptyService", false);
  ros::Rate loopRate(20);

  while (ros::ok()) {
    // Call the service
    ROS_INFO("Calling empty service.");
    std_srvs::Empty emptyMsg;
    serviceClient.call(emptyMsg);
    loopRate.sleep();
  }
  return EXIT_SUCCESS;
}

This is the server code:

#include <ros/ros.h>
#include <std_srvs/Empty.h>


bool serviceCallback(std_srvs::Empty::Request& req, std_srvs::Empty::Response& res) {
  // Does nothing, returns true.
  ROS_INFO("Empty service callback.");
  return true;
}

int main(int argc, char** argv) {
  ros::init(argc, argv, "LeakyClientNode");
  ros::NodeHandle nh;

  // Create a service callback
  ros::ServiceServer serviceServer = nh.advertiseService("EmptyService", &serviceCallback);

  ros::spin();
ROS_INFO("SHUTTING DOWN.");
  return EXIT_SUCCESS;
}

Relevant section of client valgrind output:

==22603== 8,280 bytes in 3 blocks are possibly lost in loss record 50 of 51
==22603==    at 0x4C2CC70: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==22603==    by 0x5E0E687: getifaddrs_internal (ifaddrs.c:402)
==22603==    by 0x5E0F1EF: getifaddrs (ifaddrs.c:825)
==22603==    by 0x4F0D945: ros::Transport::Transport() (in /opt/ros/indigo/lib/libroscpp.so)
==22603==    by 0x4F1592F: ros::TransportTCP::TransportTCP(ros::PollSet*, int) (in /opt/ros/indigo/lib/libroscpp.so)
==22603==    by 0x4F22CA5: ros::ServiceManager::createServiceServerLink(std::string const&, bool, std::string const&, std::string const&, std::map<std::string, std::string, std::less<std::string>, std::allocator<std::pair<std::string const, std::string> > > const&) (in /opt/ros/indigo/lib/libroscpp.so)
==22603==    by 0x4F02E6E: ros::ServiceClient::call(ros::SerializedMessage const&, ros::SerializedMessage&, std::string const&) (in /opt/ros/indigo/lib/libroscpp.so)
==22603==    by 0x40234D: main (service_client.h:102)
==22603== 
==22603== 17,746,800 bytes in 6,430 blocks are definitely lost in loss record 51 of 51
==22603==    at 0x4C2CC70: calloc (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==22603==    by 0x5E0E687: getifaddrs_internal (ifaddrs.c:402)
==22603==    by 0x5E0F1EF: getifaddrs (ifaddrs.c:825)
==22603==    by 0x4F0D945: ros::Transport::Transport() (in /opt/ros/indigo/lib/libroscpp.so)
==22603==    by 0x4F1592F: ros::TransportTCP::TransportTCP(ros::PollSet*, int) (in /opt/ros/indigo/lib/libroscpp.so)
==22603==    by 0x4F22CA5: ros::ServiceManager::createServiceServerLink(std::string const&, bool, std::string const&, std::string const&, std::map<std::string, std::string, std::less<std::string>, std::allocator ...