[libvirt] libvirt quietly dies (on keep-alive error handling?)

Hi all, we detected a floating bug of 1.0.2 version. Libvirtd mysteriously and quietly dies (without core dumps or entries in dmesg/syslog) with the following log messages. If you have any ideas of what it might be and how to deal with it - please give a hint. Many thanks. Here is the tail of libvirtd.log: 2013-04-30 11:47:46.020+000053093: info : qemuDomainUndefineFlags:5784 : Undefining domain 'vm010-002-048-002' 2013-04-30 11:47:46.195+000053087: debug : qemuDomObjFromDomainDriver:213 : Domain not found: no domain with matching uuid '52ab57f1-e198-9bd7-6ef8-74b9d34e7b03' 2013-05-01 21:00:30.997+000053085: warning : qemuDomainObjTaint:1376 : Domain id=170 name='vm010-002-060-007' uuid=41acb52e-96c0-0fe6-4d63-39d6560bc882 is tainted: custom-monitor 2013-05-02 18:37:10.430+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x88d200 after 5 keepalive messages in 30 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x8858e0 after 5 keepalive messages in 31 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x88e940 after 5 keepalive messages in 32 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x897410 after 5 keepalive messages in 32 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x8901c0 after 5 keepalive messages in 32 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x894c50 after 5 keepalive messages in 33 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x884ef0 after 5 keepalive messages in 32 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x87df20 after 5 keepalive messages in 31 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x8832e0 after 5 keepalive messages in 32 seconds 2013-05-02 18:37:14.251+000053084: warning : virKeepAliveTimerInternal:141 : No response from client 0x87dc40 after 5 keepalive messages in 32 seconds 2013-05-02 18:37:14.251+000053084: error : virFDStreamRemoveCallback:83 : internal error stream is not open 2013-05-02 18:37:14.681+000053091: error : virFDStreamRemoveCallback:83 : internal error stream is not open ====== end of log ===== 2013-05-02 18:47:28.464+0000: 97308: info : libvirt version: 1.0.2 -- wbr, Igor Lukyanov

On 05/02/2013 01:03 PM, Igor Lukyanov wrote:
Hi all, we detected a floating bug of 1.0.2 version.
Can you retest with the just-released 1.0.5? There have been several (very nasty) race bugs fixed in the meantime, including some where undefining a domain could crash libvirtd. -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org
participants (2)
-
Eric Blake
-
Igor Lukyanov