(fleet/dashboards) fix some dashboards and add new ones #387
Conversation
fbegyn
changed the title
There was a problem hiding this comment.
Comments including screenshots are in this thread on slack: https://lsstc.slack.com/archives/C05JLDDAPAR/p1715360161601749
|
The commit message also contains a typo: "dashboars" |
fbegyn
changed the title
|
I've cleaned up a lot of dashboards to bring them back into working state I believe. The only 2 broken parts are the |
The grafana dashboards are using specific datasources in the first design. This causes the UUID to break when importing these through provisioning files into other grafana instances. The fix for this is to use a datasource variable. Signed-off-by: Francis Begyn <[email protected]>
fbegyn
force-pushed
the
IT-5331-fix-snmp-dashboards
branch
from
4343112
to
a2ef94f
Compare
There was a problem hiding this comment.
The interface status plots don't render unless I scale my browse display size down to 80%. At that point they render but the interface names are illegible. E.g.
There was a problem hiding this comment.
This looks OK. Is there a way to control the order in which grafana sorts lists?
There was a problem hiding this comment.
I have to scale my browser fonts down to 80% to get this to render. The device names are sort of legible but its difficult to read and we have a lot of more switches to add.
There was a problem hiding this comment.
This seems to only be working to list the number of OSDs on a host.
There was a problem hiding this comment.
The mds_hosts and rgw_hosts drop downs don't have any values.
There was a problem hiding this comment.
The physical device metrics aren't working for me on ruka and I don't understand why. The metrics are present in prometheus.
There was a problem hiding this comment.
I have to scale the fontsizes way down to get remove horizontal scroll bars from the latency tables.
At 60% the values become visible:
There was a problem hiding this comment.
I have completed inspect all dashboard within this PR and made comments where I have concerns. WRT the Ceph dashboards. If the titles are direct copies of upstream dashboards, I think its OK to not rename them to have "Ceph" in the title but I think they need to be in a Ceph (or Rook) folder to logically group them together.
There was a problem hiding this comment.
The pool overview table is is hard to use. There are a lot of columns of duplicated or unimportant data that can be removed. The important information is on the right side of the table. The vertical height should be increased as we have a few clusters with a fair number of pools.
There was a problem hiding this comment.
This dashboard appears to be broken on both ruka and ayekan.
There was a problem hiding this comment.
This dashboard seems to be broken on ruka and ayekan for all clusters.
There was a problem hiding this comment.
We don't currently have rgw zone replication enabled on any cluster. We have tested it in the past and will probably use it later this year. It currently isn't possible to test with dashboard.
There was a problem hiding this comment.
This dashboard seems to be broken on both ruka and ayekan.
There was a problem hiding this comment.
The tables on this dashboard are difficult to read with my browser fonts at 100%. I have to scale it down to 60% to get all the data to render without horizontal scroll bars.
100%:
60%:
There was a problem hiding this comment.
This dashboard seems to be broken on both ruka and ayekan for all cluster / rgw_server combos I tried.
…han cluster labels
Use the correct key for the secret instead of the older one.
|
@KrisBuytaert @fbegyn What is the status of this PR? Is it ready for re-review? |
The grafana dashboards are using specific datasources in the first design. This causes the UUID to break when importing these through provisioning files into other grafana instances. The fix for this is to use a datasource variable.