A new filter is available on menu: filtering on URL templates that were defined for statistics. Later, I may build something more powerful to define URL templates, name them etc... But for now, this can be useful!

Seems like nothing, but I needed big rework on this part. This is not much used, but was needed for consistency and a technical debt dated back when selection of multiple Whisperers became possible.

It allows you to see when a Node has been seen by a Whisperer, and this Node's name as registered by the Whisperer. It can help understand bugs in Spider aggregation algorithms, and allow you to correct names.

Search history was available on search when the menu was reduced, but not when it was expanded. This was annoying!

That's fixed :)

Right now, it keep the last 20 manual entered searches. I can make it configurable if need be. Tell me!

I saw some people doing as me screenshots of sequence diagrams, and are forced to join the parts manually. I thought I could do something ;) !

So, from now on, you can export your currently displayed sequence diagram in SVG or PNG format with one click on a button! Convenient!

Please note that you still have to load all the flows you want to seebeforehand (using the infinite scroll). Since there is not any more autorefresh on sequence diagrams, you can scroll a long time.... ;)

Enjoy!

Automatic refresh has been there from some time, but was buggy. It is fixed now:

• Changing the settings actually does what was expected ;-)
• Sequence diagrams are not refreshed automatically, as this was annoying!
  • But they are still refreshed when you manually click on refresh button
• Grid is refreshed only when your time selection includes at least current time - 15 minutes. But that's the last time when real time data can be refreshed (second processing pass for delayed packets).

Due to new Stats panel more powerful... I removed Host stats... Moreover, it was not consistent with the rest of the U/X.

I'm sad, there were nice chunks of code there!

When clicking on an aggregated replicas Node on map, now the details panel opens with all aggregated nodes.

Quick filter have been added on client identification tooltips:

They work with Ctrl and Shift for negative filtering and replace filtering. Enjoy!

Stats panel is great ! But stats are taking too much space, and it was difficult to keep track of what stat existed. So I added a statsBar (like a task bar ;-) )

At the top of the stats panel, it tells you what stats config exists and allows you to:

• Delete a stat
• Jump directly to a stat
• See full stat description in tooltip

Moreover, when creating a new stat (empty or as a detail of a pivot table), you are automatically scrolled to it.

This has been awaiting for long, and I wasn't sure what I'll come out with. The result has gone far from the first design... it does much more :-) !

• Dynamic statistics, linked to the full power of the filters already existing
• Possibility to compare stats for different selection
• Possibility to do statistics over the full selected time windows, or with a moving period inside this window
  • Avg duration over 2h
  • Avg duration by s over 2h are not the same stats
• Possibility to use stats to dig into analysis
• And cherry on the cake, possibility to use aggregations and have a pivot table to have a refined view of the system
  • That one was tough for my brain to build ;-)

I hope you'll help me increase the potential of this part with your ideas and comments.

Example 1: Response time analysis

Response time statistics for POST and PATCH verbs on SIT1 this february 28th.

More details on the POST 200 on frenchFines, that obviously has an issue

We can see that the issue is only with our own tests (technical admin user) !

Zoom on the statistic (with the magnifying glass), and you have access to heat map, statistical distribution of the metric, and more figures.

For our case, we notice that the bad response times are all focused on specific timeset. Then, selecting any of the graphics area, you can jump to list view, and get all details needed!

Example 2: Server load

Statistics of count of requests by minute for Fines. We can see the maximum at 200 request by minutes that was performed by the technical admin.

In fact the bad response times where do to when integration team tested in non regression the concurrency issue. No need to alarm.

Example 3: Stats by URL templates

One other key features is the possibility to get statistics not by URL, but by URL templates, in order to get REST endpoint statistics! Ex:

URL templates need to be configured in Settings panel.

 

What's your feeling? Killing feature or not? :) Play with it and tell me what you think... and if you find bugs!

 

Note: For safety, I limited the number of distinct terms (URI, Servers... returned in the aggregations):

• Clients: 50
• Servers: 50
• URLs: 25
• Status codes: 50 (should be enough)