libvirtd might abort when virsh is run repeatedly under extreme loads.
Issue
-
libvirtd might abort when virsh with either of the following options is run repeatedly on a host and guest which load is extremely high due to Unixbench.
- nodeinfo
- domuuid
- domid
- dumpxml
- domstate
-
virsh run under the above condition aborts with the following error message.
# cat /var/log/messages
Jun 15 16:20:37 tx300s4-2 abrt[4797]: saved core dump of pid 2262 (/usr/sbin/libvirtd)
to /var/spool/abrt/ccpp-1276586437-2262.new/coredump (98840576 bytes)
Jun 15 16:20:37 tx300s4-2 abrtd: Directory 'ccpp-1276586437-2262' creation detected
Jun 15 16:20:37 tx300s4-2 libvirtd: Could not find keytab file: /etc/libvirt/krb5.tab:
No such file or directory
Jun 15 16:20:37 tx300s4-2 abrtd: New crash /var/spool/abrt/ccpp-1276586437-2262,processing
-
Step to Reproduce:
- Make a host load as high as possible.
- Make 4 guests load as high as possible.
- Run all of the following commands repeatedly on each guest.
# virsh nodeinfo # virsh domuuid <domain-name-or-id> # virsh domid <domain-name-or-uuid> # virsh dumpxml <domain-id> # virsh domstate <domain-id>
Environment
- Red Hat Enterprise Linux 6
- Architecture: x86_64
- Kernel Version: kernel-2.6.32-30.el6.x86_64
- Related Package Version: libvirt-0.8.1-7.el6.x86_64
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.