This is a new issue, though, and I haven’t seen any documentation about a change in how recursive workflows work. Surely it can’t be an intended change?
Hi all,
The team is investigating these reports. Just to make sure, have you all submitted a bug report via the Bubble Support Center page ?
I got this response from Bubble, which is completely unacceptable:
“It’s crucial to ensure that backend workflows are not running concurrently with other resource-intensive processes, as this can result in timeouts. One strategy to mitigate this is to restructure workflows by breaking down heavy actions or introducing delays between them.”
The backend workflow that previously took just 10 minutes is now taking at least five hours—if it even completes at all. And yet, they’re suggesting I introduce even more delays instead of addressing the root cause of the issue. This is beyond frustrating and does not provide a real solution.
Backend workflows timeout after 5 minutes. Are you talking about a recursive workflow? If so, which actions are inside of it?
The context is 140WU per run. So 5600 runs cost me 700k WU. Setting a recursion limit of say 2000 will break many other things. So basically I now have to break the process into enough steps to make this 50 000 - 100 000 runs so I can effectively use recursion protection. When you look at the way WU is calculated we’re looking at 30% premium probably.
In my case, I’m running the backend workflow on a short list (6 items, for example). The problem is not in the size of the list.
(As demonstrated on the video above)
For what it’s worth, it does look like there’s some behaviour change in backend workflows since Feb 26th, which was also when there was database migration, but if you share more details about the workflows you’re having problems with (screenshots of each step), and share the logs, then you can get a resolution here.
Without sharing any logs or details of the workflow, you won’t get anywhere trying to resolve it.
This is a normal API workflow on a list of things. The list has about 1500 entries. I’m simply looping through each entry and updating one field.
Yes, and what do your logs say? At what point does it fail? What do the logs say when it fails?
Exactly! This is clearly an issue on their end. I’ve had to cancel and restart the process multiple times, and now they’re even charging me for the additional workload…
George,
Here a video showing all the workflows and the logs:
As you can see on this manually created log, when we use List: #item number, after a few iterations, it’s not retrieving the data:
For the first 2 items, it retrieved the information “Category”. After that, it did not.
1 Like
The logs show that the processing is happening but each time I’ve had to cancel because it was moving extremely slowly. This time, I’m going to let it run to see if it even completes. So far, it’s been about two hours - yet the same size batch used to process in 10 minutes or less.
I also have one backend workflow that is really acting up, noticed it this Monday. All without any changes in it or anything connected to it. Has been running since Aug last year without issues.
I have submitted a bug report, and really need bubble to act on it as fast as possible.
1 Like
What happens if you modify the Schedule action such that it runs only when IterecaoTotal > IterecaoAtual and Result of step 10 is not empty?
Our first reported issue occurred on Monday as well.
The API process finally finished after 3.5 hours, but I noticed that it randomly skipped some entries. This is completely unacceptable!!
From my perspective, this is a complete system failure…
1 Like
It didn’t work as well. It’s something on their side, for sure.
What shocks me is how long this is taking to be fixed (5 days). There must be thousands of apps affected and the registers are not being created of modified properly. We should at least be informed about what is happening and when everything will come back to normal.
This is an extremely serious problem, we need your help as fast as possible.
2 Likes
I can’t even wrap my head around the extent of the data that’s been skewed so far, or how anyone is supposed to track down what’s been missed. This could be a nightmare to untangle.
One more quick check to make sure everyone in this thread has submitted a bug report in our support page Bubble Support Center
It’s the best way to increase visibility with the team so they can triage any potential errors
I have submitted a bug report, but they’ve just kicked it off to a “workflow” team. It is frustrating that Bubble has not acknowledged the issue or provided updates on it. Many of us are affected and know it; I suspect many others are and just don’t know it.
1 Like