Hi,
I employed libvirt to monitor 50 domains in one node(host). my
monitor calls libvirtd for every stats such as cpu, memory, disk every
10 sec. All my requests are made in one thread.
And I found that the libvirtd may cost 30% or above cpu time. Is it
normal in my case?
I tried to reduce this cpu consumption and found some variable in
libvirtd.conf and got a little confused by variables like max_clients,
max_workers, max_requests.
In my opinion, max_clients limits the number of virConnect created by app.
max_workers limits the number of thread to handle requests from
clients. Does that mean requests from one client may be handled in
multi-thread way if max_workers is greater than max_workers?
max_request limits the concurrent RPC calls according the comments
in libvirtd.conf. I wonder if it can improve the response time if I
give a higher value in the case that all my requests are made in one
thread?
I've tried several different configurations and they didn't reduce
the cpu usage effectively.
_______________________________________________
libvirt-users mailing list
libvirt-users@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/libvirt-users
