Hi Nick,
I know what you mean, no matter how hard you try something unexpected
> always happens. That said I think OSD timeouts should be higher than HSRP
> and spanning tree convergence times, so I think it should survive most
> incidents that I can think of.
>
So for high speed networking (40+
Hi ,
yes librbd can use a lot of cpu, mainly if you want to do a lot of iops.
I have benched it vs krbd driver, and even with tcmalloc/jemalloc fio librbd use
twice more cpu than krbd.
as qemu use only 1 thread by default for all disks, this is a bottleneck.
using multiple iothreads (1by disk)
> -Original Message-
> From: Alex Gorbachev [mailto:a...@iss-integration.com]
> Sent: 28 June 2015 18:57
> To: Nick Fisk
> Cc: ceph-users@lists.ceph.com
> Subject: Re: [ceph-users] Redundant networks in Ceph
>
> Hi Nick,
>
> I know what you mean, no matter how hard you try something unexp
