Are there Bugs in WU metric reporting?

October 1st is fast approaching…

I’m running simple tests. I have a page, with a single Repeating Group. It does a simple search of a data type. It has one constraint on it that is connected to an option set value, which is a field on the data type being searched.

When I look at the WU metrics, after loading the page ONCE, I see some pretty strange reporting of WUs, at least it seems strange to me, but of course the logic I follow is not always the same logic Bubble follows.

What I have to begin is page load and Fetching Data - to be expected


Then when I click into the Fetching Data I see 3 different types of Searches

  1. Search
  2. Aggregate Search
  3. Real Time Search

When I click into each, I see that each one is attributed to the same search (the only search performed on the page which was loaded ONCE)



Based on the WU activity pricing in the manual I am assuming that Search is two components

Which makes sense as a total value of 0.37 WUs reported, which could be the 0.3 + (X number of characters *0.000003).

Then should I assume that the Aggregate search of 0.20 WUs is ‘Performing an Aggregate Query in the database’? And why would that not have just been the ‘Performing a database search’?

And, what is an Aggregate Query in the database? Is that supposed to be a simple “Do a Search”? Something like below, which is what I am testing.

Screen Shot 2024-08-14 at 2.10.10 AM

And why was it showing the Real-Time Search as 0.01 WUs? In the manual it shows as a cost for ‘each call to watch for real time updates’ as 0.005 WUs. Keep in mind, I only ran this search ONCE, as I only loaded the page ONCE, and the search as nothing that would have caused it to have been called to watch for a real time update anymore than just the ONE time the search was performed.


Lastly, why would there be no charges for ‘each thing returned from the database’ on this simple search?

The search results had a count of 22,000 or so records, however, the repeating group was set to fixed number of rows at 4, so the assumption there is that only 4 things were actually returned from the database…but even so, that would be 4*0.015=0.06 WUs

All in all this was 0.72 WUs reported in the app metrics.

I would have expected from the manual a charge of 0.785 if I was being generous and stated I expected to be charged the 0.20 WUs for an aggregate search, but I don’t, so in reality I expect that I’d be charged:

0.15 Page Load
0.3 Search
0.07 all characters returned from database
0.005 for each call to watch for real time updates (just ONE)

Total of 0.585 WUs

Expect a mess after October 1, 2024 - likely a nightmare of a few months, causing delays in bugs getting fixed, rollouts of expected features and delays in delivering improved performance and reliability.

@fede.bubble could you reach out to somebody at Bubble who is in charge or works with the team on WUs and the metics for them to reply in on this thread to help shed some light on what is going on. I’ve been around Bubble for a bit and I’m sort of lost on why the numbers just don’t add up as expected.

1 Like

From what I know, aggregate is triggered when, for example, you have :count (take note that, as mentionned in another post, a = 0 for example, will trigger an msearch instead of maggregate. From what I know, this is not fixed…). You can see in network tab maggregate, so I guess this is the aggregate search.
a search should trigger msearch. I don’t see any reason in your example for maggregate actually. But maybe it’s somewhere else.

1 Like

Absolutely. Could you send me a DM with the app details on where this is coming from? That’ll make it easier for the team to take a look.
Btw, success is pretty good at these kinds of questions too.

3 Likes

For me this is not a success at all if we need to contact support to understand WU…

Pretty sure I noticed sometime ago during testing, a realtime search gets done once after the RG is loaded and then subsequently everytime that datatype gets updated in the DB.

Hence why i don’t use searches as data sources unless I absolutely need realtime updates to a list.

I had a text element that referenced the repeating group list of things count…that on it’s own should not incur any WUs as it is referencing values already on the page, but I suppose that is a bug that is not fixed.

No updates took place. Must be a bug that is causing extra WU costs that should not be incurred.

What do you do instead? Use workflow action for search and then display list in RG?

I just store everything in states. It’s easier to manage since I don’t have to crack my brain trying to look for these things.

I have a feeling that’s the case. Minor, but it can add up. Though I really hope we’re wrong.

1 Like

This topic was automatically closed after 14 days. New replies are no longer allowed.