Am 13.07.2013 17:02, schrieb James B. Byrne:
>
> On Fri, July 12, 2013 13:34, Alexander Dalloz wrote:
>> Am 12.07.2013 16:06, schrieb James B. Byrne:
>>>
>>> CentOS-6.4_x86_64
>>>
>>> I have a KVM virtual machine (several actually) which acts as a warm
>>> spare for another kvm instance running on a different physical host.
>>> Each hour it backs up the live system using rsync. This morning I
>>> discovered that the backup instance was running at 99% utilisation
>>> and that the process consuming the entire virtual cpu was this:
>>>
>>> sh -c rpm -q -a --queryformat
>>> "%{NAME}\n%{VERSION}-%{RELEASE}\n%{EPOCH}\n%{GROUP}\n%{SUMMARY}\n\n"
>>> 2>/dev/null
[ ... ]
> Does anyone have any suggestion as why this process would get hung up
> like this?
Probably you had updates installed on the primary VM, then your rsync
job did the syncronization while at the same time the rpm job had
started on the secondary system, thus the database query (rpm_db is in
fact a database) got confused and failed to operate properly. That's to
be expected if you work with a database and inject changes to its files
with something like rsync.
Alexander