Repository navigation
Conversation
|
I'm going to have to stew on this one. It's a mass change in the middle of the game. May have negative impact on the community. I'll give it some thought next week. In the mean time, you can fix the issues from the previous pull which is much more benign. |
|
That’s fair, I understand the concern about changing existing graphs and templates across the community. For context, I’ve been running Cacti in an enterprise environment for a long time, and this has been a recurring source of confusion for management and other users. They see the inbound/outbound peaks on the graph and naturally expect the “Maximum” values underneath to match. I’ve regularly had to explain that the existing Maximum represents the maximum of the averaged series, while these templates also display a MAX consolidation series. My aim is to make that distinction clear within the graph itself, particularly when it’s exported, screenshotted or included in a report without someone there to explain it. I’m happy to discuss a less disruptive approach, including making the changes opt-in rather than automatically updating existing graphs. |
e63f128 to
830eaf0
Compare
830eaf0 to
50e01b8
Compare

The standard interface traffic templates draw MAX traffic lines alongside averaged traffic, but the legend’s “Maximum” reports the highest averaged value. This can make a visible spike appear inconsistent with the legend.
This renames that statistic to “Max average:” and adds “Max peak:” directly underneath it for each direction, with aligned labels and values. The new reading follows the corresponding MAX series. Bits/sec, bytes/sec, bandwidth-total and 95th percentile variants are covered; existing totals, units and percentile definitions are preserved.
Existing graphs are included: the upgrade identifies the five stock templates by stable hashes, updates their linked graph items and adds the template input associations needed by future graphs. The same migration runs after installation package imports, leaving the signed packages intact. Running it again does not add duplicate peak readings. Layouts that no longer match the stock legend are left unchanged.
I tested the layout on 69 traffic graphs on my test setup and checked the rendered daily and weekly views. An isolated migration regression check passed for all five upstream template definitions and five existing graph copies using different database IDs, including preservation of calculations, template/input linkage and repeat execution. PHP syntax checks passed. The automated migration has not yet been tested against a separate full database upgrade.
The migration is connected to the 1.2.32 upgrade hook and installation package imports. This change does not depend on the two-year panel change.