dlmarion commented on issue #4973:
URL: https://github.com/apache/accumulo/issues/4973#issuecomment-2414541706
I'm not a proponent of the Monitor as I think there are other things well
suited. In fact, I was maybe the loudest to say that the Monitor should go for
4.0. However, after giving this some more thought:
1. There is a real-world cost to setting up, running, and maintaining
metric and log aggregation and display systems. For large installations, doing
this likely makes sense. For smaller systems, or in cases where someone is
trying to evaluate whether Accumulo is right for them, this could be a negative.
2. One thing that I think the Monitor does well, is condensing a lot of
information into tables. If you look at the Grafana Table
[documentation](https://grafana.com/docs/grafana/latest/panels-visualizations/visualizations/table/#supported-data-formats),
for example, its supported formats are csv-like or a transformation on some
query. The ability to do this via query, and the syntax used, would likely be
dependent on the metric database being used. It's quite possible that a Table
that we have on the Monitor today can't be built in the same way in Grafana or
other monitoring systems.
3. I have been putting some thought into how we might change the Monitor
to display some of the new concepts that we have introduced, but using the
Metrics that we generate instead of private information gathered via internal
code that is funneled back to the Monitor via the Manager. If we did this in a
kind of eat-your-own-dog-food type of way, then we could annotate (in a
tool-tip?) which metrics were being used for the graph or table, so that a user
trying to implement their own solution didn't have to guess which metrics made
sense to go together. We should also be able to determine which metrics are
not really that useful, and which ones need to be added to create a new graph
or table. In this case, the Monitor is the exemplar, and if it breaks down at
some scale, so be it. We have given the user the documentation they need to
build their own.
Regarding Monitor changes:
1. The External Compactions page needs to be removed, these are just major
compactions now.
2. The main page might just be a summary of the resource groups. Type and
number of server processes, how many are configured, active, and idle
3. We should likely have a page for each resource group
4. We likely want to remove OS-type things (e.g. OS load)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
