Kyle Bader

The bigger problem with cgroups is that they only provide independent limits for read and write, compared with libdrbd/virtio which provide the ability to express limits against the aggregate of...

Probably would need something like this KEP to do accounting - https://github.com/kubernetes/enhancements/pull/1353 Basically if you know a cluster can provide 100k iops and 100TB, then you need to add up...

You can "fix" this by enabling CORS on the bucket created with boto. Perhaps when you use the web interface it does this automatically when it creates the bucket. Is...

They don't allow scheduling at a specific time, they allow specifying a window in which scrubs can take place. I'm not sure these will work for us.. osd_scrub_begin_hour, default 0...

The idea here would be to disable collectl, and rely on Calamari for collecting metrics during test runs. We could then use graphite's graph composer to quickly analyze the results...

I tested this, and it's awesome. I was having wild variance between test iterations when the first iteration would create a ton of objects. If I run the same test...

In the librbdfio benchmark we have use_existing_volumes. The problem with use_existing_volumes is it doesn't create the volumes if they already exist. Ideally there would be an option where it creates...

Well I don't think we should alert if the counter on the host is higher than some threshold, since the counter would be total drops since the machine was booted....

Probably should be done through zone group hostnames - https://docs.ceph.com/docs/luminous/radosgw/multisite/#set-a-zone-group

Ceph merged a ListObjecstV2 support over a year ago in this [PR](https://github.com/ceph/ceph/pull/26512). The Ceph [tracker issue](https://tracker.ceph.com/issues/37801) linked above was worked around by the reporter by using ListObjects, once they noticed...