Bug #4523
closedosd: read stats not updated
0%
Description
A week ago I already added to http://tracker.ceph.com/issues/2209#change-18705 but as this bug is still closed I assume it got missed and so I create this new issue.
My cluster does not seem to update it's read stats in argonaut [0.48.2-0ubuntu2] and latest bobtail [ceph version 0.56.3 (6eb7e15a4783b122e9b0c85ea9ba064145958aa5)].
Both clusters only report a changing wd, rd always stays at the same very low (initial?) value.
Output of "rados df" from my bobtail cluster: http://pastebin.com/Ehc1cKyL
Output of "ceph -w" from my bobtail cluster: http://pastebin.com/A9f2012X
I only use qemu rbd images, could it be related to this? There are around 100 vms running, so there are a lot of reads going on for sure.
Output of "rados df" from my bobtail cluster a few minutes later for comparison: http://pastebin.com/C0nrf02T