Updated the default Grafana time window to last 30 minutes and set the default screen refresh rate to 30s

[butson]

Here are my latest dashboards. there have been several changes see comments in commit e6617bb.

If you wish to preview these changes before approving the request. You can:

git clone git@github.com:nuodb/nuodb-insights.git
cd nuodb-insights
git checkout dbutson/dashboards

docker run -d --name grafana -p 3000:3000 --env INFLUX_HOST=monitor.don.mynuodb.com -v $PWD/conf/provisioning:/etc/grafana/provisioning grafana/grafana:latest
browser http://localhost:3000

I don't know how long monitor.don.mynuodb.com will be running but I assume during this review period. Switch time frame if test runs are not currently running. Use 2/16 and DIRK database should give you some data.

[acabrele]

adhoc.json has ENOVIA saved as current value

[sivanov-nuodb]

Thanks a lot for adding descriptions! This is a great improvement.
I've added some minor comments about the visualization.

5. Annotation now based upon database name tag. So looking at a
   different database you can filter which annotations to look at.

• only works with one database, not sure if we could make toggle be
  just to enable/disable and based set of annotations based up all
  selected databases.

Annotation query filtered by template variable doesn't seem to work correctly with multi-value filters. For example, if you have multiple database selections or "All", they are not shown.
I found it working with multiple databases selection if you switch on "Match Any" for the annotation tag query. To work with All database value, the annotation should be marked with .* tag, which is not intuitive.

[sivanov-nuodb]

Here it says "PercentSystemTime" but the query below selects "PercentUserTime"?

[butson]

looks like a problem. there looks like there are more problems with dashboard. As often measurement name does not match query.

[sivanov-nuodb]

From the Grafana UI this looks okay. I guess you fixed it or it was a scrambled json output.

[butson]

I've merged changes into the branch , please review if you wish.

As for how annotations are working, we need to decide what we want, I believe we are limited in what we can do. What I currently have attempted to do is use the database name as a tag to annotation relative to that database. You would only see annotations for that database if only that database is selected in template variable db. So it will not work with multiple databases / ALL. As for default annotations. we could enable, enable with a global tag to use or disable. I have not changed all dashboards (such as nuodb-ops-overview) to deal with annotations.

Please comment on what you believe is correct. or we open a request following this review to fix the annotations.

[acabrele]

I have not changed all dashboards (such as nuodb-ops-overview) to deal with annotations.

OK, this is why I see a difference in behaviour. It would be good for them all to be consistent.

Please comment on what you believe is correct. Or we open a request following this review to fix the annotations.

From a usability perspective the only issue I see is disappearing annotations when you don't add a tag, but once you know... you know. If multiple DBs are on a single domain / host(s), global option could be useful too. Or, could just use the .* as Stanimir commented.

[sivanov-nuodb]

It's hard to see if all aliases are consistent. i.e SM|TE(db: $tag_db, nodeid: $tag_nodeid, startid: $tag_startid).

Other than this the changes look good to me. Thanks!

[mkysel]

Here the collection of changes

1. Description added to nuodb dashboards.

2. Current values in nuodb dashboards better defined as last 1 minute
interval.
3. Filter variables are computed differently to take time range into
consideration.
4. Default time window last 30 minutes and refresh rate set to 30s.
- unfortunately you can't control refresh rate based upon time window.
5. Annotation now based upon database name tag.  So looking at a
different database you can filter which annotations to look at.
- only works with one database,  not sure if we could make toggle be
just to enable/disable and based set of annotations based up all
selected databases.
6. Removed connections as it really did add any additional value over
what we have with other dashboards.
7. threads left as is for now,  but it's not correct.   Would like to
change collector to get more info such as (engine type, startid and
nodeid) from proc cmdline.