Hi Bairo, On 12.09.2017 00:41, Blair Bethwaite wrote: > On 12 September 2017 at 01:15, Blair Bethwaite > <blair.bethwa...@gmail.com> wrote: >> Flow-control may well just mask the real problem. Did your throughput >> improve? Also, does that mean flow-control is on for all ports on the >> switch...? IIUC, then such "global pause" flow-control will mean switchports >> with links to upstream network devices will also be paused if the switch is >> attempting to pass packets from those ports down to a congested host. > > Sorry, I didn't finish explaining that... The problem in the above > scenario is that global-pause is link-level, so even if you have other > flows destined for non-congested hosts on the same switch and plenty > of upstream/fabric link capacity for them, those flows will also get > paused due to a single congested downstream host. For any particular > topology and congestion rate there is likely to be a tipping point.
Flow control was activated only on the 25 GBit/s ports. Those ports are dedicated ports for ceph and traffic will not leave the switch. I used a seperate vlan. Regards, Andreas _______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
