therm
121 Posts
Quote from therm on August 21, 2017, 6:42 am
Hi,
could you please explain what exactly is measured by the dashboard chart throughput? My understanding is that it takes 'ceph -w'-output.
If so, how are writes ment? If 100MB are written, does it mean that there were about 100MB Client IO written to the cluster (and 300MB because of replication factor 3) or were there about 33 MB written and ceph shows us the real traffic including replication traffic?
Thanks in advance,
Dennis
Hi,
could you please explain what exactly is measured by the dashboard chart throughput? My understanding is that it takes 'ceph -w'-output.
If so, how are writes ment? If 100MB are written, does it mean that there were about 100MB Client IO written to the cluster (and 300MB because of replication factor 3) or were there about 33 MB written and ceph shows us the real traffic including replication traffic?
Thanks in advance,
Dennis
admin
2,930 Posts
Quote from admin on August 21, 2017, 12:30 pm
It is the first: 100MB /s by client which will internally use 300 MB/s
We use the value in
osd pool stats --format json
for write_bytes_sec and read_bytes_sec
it is the same as ceph -w
It is the first: 100MB /s by client which will internally use 300 MB/s
We use the value in
osd pool stats --format json
for write_bytes_sec and read_bytes_sec
it is the same as ceph -w
