Fedora Account System
Red Hat Associate
Red Hat Customer
Created attachment 1350126 [details] Bricks dashboard Description of problem: The bricks dashboard has 2 panels for Throughput with differing output. It's unclear what the difference is at first glance. Either one of them needs a different label, or we should only have 1 of them. It would be also good to describe what they are. Version-Release number of selected component (if applicable): How reproducible: Click on Bricks dashboard and any brick, and you'll see 2 dashboard panels called "Throughput." Steps to Reproduce: 1. Click on Bricks dashboard and any brick, and you'll see 2 dashboard panels called "Throughput." 2. 3. Actual results: 2 "Throughput" panels in the Bricks dashboard but mean different things but unclear what it is when looking at it. Expected results: Having a single Throughput panel. Additional info: See https://github.com/Tendrl/monitoring-integration/issues/251.
I would propose to take this in a future release as we are close 3.3.1 timelines and this bug not critical enough to affect the core functionality. We can make sure that the this is captured clearly in documentation.
From looking at monitoring-integration/etc/tendrl/monitoring-integration/grafana/dashboards/tendrl-gluster-bricks.json, here is what I saw: (1) the first Throughput panel is showing the sum of Brick IOPS (read + write) for the brick path tendrl.clusters.$cluster_id.volumes.$volume_name.nodes.$host_name.bricks.$brick_path.iops.gauge-read + tendrl.clusters.$cluster_id.volumes.$volume_name.nodes.$host_name.bricks.$brick_path.iops.gauge-write Current Description: N/A (2) the second Throughput panel is showing disk read and disk write based on tendrl.clusters.$cluster_id.volumes.$volume_name.nodes.$host_name.bricks.$brick_path.device..disk_octets.read and tendrl.clusters.$cluster_id.volumes.$volume_name.nodes.$host_name.bricks.$brick_path.device..disk_octets.write. Current Description: "The Throughput panel displays brick-level read and write operations fetched using “gluster volume profile” Perhaps calling the first Throughput panel what it's showing i.e. "IOPS" with a description of "The IOPS panel displays the sum of IOPS (read + write) for a given brick path". Have we confirmed that we actually collect brick path IOPS data? I've not yet seen this data yet. If no, then we probably should not show it. Can someone verify please?
Since RHGS 3.4.0 (at least since tendrl-monitoring-integration-1.6.3-7.el7rhgs.noarch), there is just single panel labeled Throughput in Brick Dashboard of RHGSWA Grafana based dashboard. This change must have landed in RHGSWA codebase via one of rebases done during 3.4.0 development.