Description of problem: When I use command gluster volume top Staging read-perf list-cnt 1, I always get 0MBps traffic even on highly loaded Gluster cluster. Output bellow: Brick: dfs01:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...hmU2oqe86e_dataset.labels/dataset.labels.csv 2012-08-28 14:55:20.481924 Brick: dfs05:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...d3b0235-28fe-4a44-a48c-e3d9480b9fde/data.csv 2012-08-28 14:55:21.923247 Brick: dfs03:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...lX6kx6JMjtbyTevC_dataset.sms/dataset.sms.csv 2012-08-28 14:55:21.637640 Brick: dfs07:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...56f0a1f-5fe5-4a32-8776-95ca6a652c7f/data.csv 2012-08-28 14:55:21.915933 Brick: dfs09:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...AphzNgYnlpzIqucEV_dataset.stories/upload.zip 2012-08-28 14:55:19.873379 Brick: dfs04:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...55ed3d6-29f1-4071-882b-aac8502275cd/data.csv 2012-08-28 14:55:21.951723 Brick: dfs10:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...0a_dataset.zendesktags/d_zendesktags_tag.csv 2012-08-28 14:55:20.897776 Brick: dfs02:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ....labelstostories/dataset.labelstostories.csv 2012-08-28 14:55:21.355501 Brick: dfs08:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ...e26e86e-44ef-4d35-b371-2a03ac7359f1/data.csv 2012-08-28 14:55:21.950289 Brick: dfs06:/mnt/gluster/Staging Throughput 0.00 MBps time 0.0000 secs MBps Filename Time ==== ======== ==== 0 ..._dataset.stories/d_stories_iteration.csv.log 2012-08-28 14:55:19.457255 Version-Release number of selected component (if applicable): Vanilla glusterfs-3.3.0 How reproducible: gluster volume top Staging read-perf or write-perf Steps to Reproduce: 1. Run gluster volume top Staging read-perf or write-perf Actual results: 0MBps traffic Expected results: Traffic by reality Additional info: Gluster setup of 10 nodes with replica 2. Clients are using same version of glusterfs and glusterfs mounts. I can provide more debug informations, however I am not sure what may help.
Have you turned the latency-measurement on? If it's not set then mbps for individual files will not be displayed. gluster volume set Staging latency-measurement on And Can you please run read-perf/write-perf with the block size and count? Like gluster volume top Staging read-perf bs 1024 count 100 or gluster volume top Staging write-perf bs 1024 count 500 list-cnt 5 Currently Throughput of the total brick will be displayed as zero until you specify the block-size (bs) and count. Can you please try above commands and get back to us?
Sorry, I have probably misunderstood this command purpose - using your parameters work. However, I haven't found command that would be able to show current traffic on whole gluster volume (or per-brick). Volume profile command is fine, but too low-level for overall statistics. Thank you for your answer.
We don't have any gluster commands to capture the overall gluster volume traffic statistics. gluster profile/top commands were introduced for getting statistics per brick level. You might want to try out (that is, if you haven't already) some other gluster top commands like open, read, write, opendir, readdir, write-perf and read-perf. You can run gluster volume help for the usage of those options.
Ok, thank you. I am closing the issue as invalid.