On 10.05.2016 22:02, Eric Wheeler wrote: > Hello all, > > We've been using libvirt for years and it works great, however we just hit > a bug. > > This is running libvirt-1.2.17-13.el7_2.4, however strangely, we had the > same backtrace on a complete difference system in a different data center > using version libvirt-1.2.17-13.el7_2.3. Both traces are pasted below. > This rarely happens, so I'm not sure how to reproduce it. I'm opening to > suggestions if you would like to try anything. This usually means that we are overwriting memory somewhere. Worse, we are overwriting glibc's memory somewhere so next malloc() we make results in this crash. Usually, what I do in this situations is to run libvirtd under valgrind. I suspect it's daemon that crashed because of all the libraries linked in. But be prepared, that doing so will decrease libvirt's performance. Also, it would be nice to see what were recent actions running on the daemon so that we can try to reproduce. Worst case scenario is that there's just one API that corrupts memory and is executed very soon and it's only later, much later that the crash occurs. At any rate, valgrind should help. Michal -- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list