Marvel 2 is missing IOPs and Network traffic graphs

monitoring

(ajay.bh111) #1

Key information trend in ES cluster nodes about network traffic, IOPs, storage used are missing and there does not appear to be any link to add additional dashboard with basic license.


Missing Metrics in Marvel 2.X compared to 1.X?
(Steve Kearns) #2

Hi Ajay,

Thanks for bringing this up. In Marvel 2.0, we're trying to focus on the highest priority metrics and making things easy to understand.

On the Nodes List, we show disk free space and on the node detail page, we show the volume of data used by ES.

Some of the other statistics, like IOPS, are no longer available from ES, because we stopped using Sigar, a C-code library that gathers system stats. We stopped using Sigar because it was causing stability issues on some platforms, and isn't actively maintained anymore.

While these aren't available from ES anymore (and arguably were never appropriate to report from ES), they are really useful statistics to have. I created an issue in Topbeat to capture IOPS and disk utilization: https://github.com/elastic/topbeat/issues/118 .

I hope that helps!
Steve


(ajay.bh111) #3

Thanks Steve.

If some stats are not available, it would be better to include option to capture custom stats from other sources on the node , keep the data in custom indices and plot them using Kibana. Systat/iostats like utilities can be integrated thru custom interfaces/plugins. These are important data when monitoring the service with in cluster and investigating performance and capacity planning issues. At the moment no provision exist to add any custom data capture from the node or ES stats. Without them Marvel be lacking critical visibility on ES cluster operations.

Ajay


(system) #4