For the third day in a row our $600-a-month app has maxed out capacity just under an hour ago. When that happened, there were 2 workflows running in that minute, 8 people on our page and the capacity graph showed us at 1.7% of capacity used in the same minute we maxed out. I’ve gone through all usage tables to review workflows, searches, page loads and I can’t find anything at all that would have caused the spike (I was thinking it could have been scheduled workflows, but those only used 3% of our capacity in the last hour.
Has anyone else come up with any consistently effective ways of diagnosing why your app is hitting full capacity and freezing for people when none of the usage statistics show it coming anywhere close?
I’ve tried to address this with Bubble directly, but all they can seem to do is point me toward diagnostic tools that all show the same thing: this app is nowhere near capacity when it suddenly and repeatedly maxes out.
Unfortunately no - I’m actually glad you posted again. I saw your post in the other thread and I’m watching this with interest. What would be super helpful is “this is the exact workflow that caused the capacity spike” - But diagnosing is very difficult.
We are currently having a discussion at my software company about taking one of our Bubble MVPs and turning it into a production candidate on a Production plan… However, what you just described plus the outage 2 days ago really is making this a tough sell.
Thank you for posting this Brian - Wish I could help. Perhaps some of the other veterans on here will have some advice.
I believe you can see users session in Google Analytics(if you have one), and since you have time when capacity has maxed out you can trace what 8 users was doing in that particular moment.
Thanks for coming back @emmanuel and @GregoryKing, @w.fly and @sat_miha for the suggestions - I’ve filed multiple bug reports with support on this one already and haven’t made much headway. We made a few capacity reducing adjustments over the past two days after @eve suggested taking a look at the capacity usage pie chart and added another $100+ a month of capacity, but we’re still capping out so we haven’t found the solution yet.
We’re a week on from the original bug report here and six days after @emmanuel asked us to forward this on to the engineering team (we already had, several times) . We’ve also provided the Bubble engineers with the helpful pointer @nocodeventure offered. And we still haven’t heard anything at all back from Team Bubble.
I know there have been questions in other threads about Bug reporting and potential improvements to the process with a couple interesting posts by @allenyang and @eve. I also know that a number of reported bugs turn out not to be bugs at all, but errors in logic, which makes the job even harder. But when things like this pop up that obviously are not related to errors in logic and aren’t things we can easily use the Bubble Editor to help resolve, I think it’s fair that we expect help…especially given that we’re paying for it.
In this specific instance, you have me (a non-coding founder) who has an app that is actually finding a level of commercial success - the growth of which is now being hindered by Bubble itself (or the network of Bubble and it’s approved plugins). I’ve been asking for help for weeks from the team that told me I didn’t need to be a coder to succeed here about issues I can’t see and don’t understand from my side.
And I’ve gotten nowhere. Which is frustrating to say the least.
Funny enough, I just got contacted by a member of my software team looking at an MVP I recently completed (legacy professional plan - 2 units). He goes - “hey, so what’s going on, I’m just going through the pages and it’s hiccuping”… Yup, sure enough, it was. I go look at capacity… hmm… no maxes, but I am high on the average. Looks like an upgrade may be in the future. lol.
Luckily this is just being used by 1 team member right now… Man, I can’t imagine if this was a client or worse, clients, who contacted me having issues - Beyond just throwing down more and more money for capacity, you’re kinda stuck it seems. I feel for ya.
Brian, I’m going to respond to your case directly to you in a minute, but one tip which might help in case people weren’t aware of this feature: in the Logs > Capacity tab, in the section with the pie chart, you can click on specific parts of the pie chart to ‘zoom in’ for the next level of breakdown. This should help you identify some potential culprits in terms of specific workflows or elements to take a look at…
For me, it’s usually hit or miss on that pie chart. For instance, in the screenshot below, “Other” is taking up a large chunk… But when clicked, I don’t know what it is.
I feel for you Brian you shouldn’t need to throw that much money at capacity! That kind of money for capacity is insane and certainly without getting immediate support. I pay Amazon Web Services $100 a month for premium support and chat or phone support is immediate.
This is so disappointing! Perhaps if Bubble offered a monthly fee for immediate support or at least a 24 hour response.
I hope you get this resolved and the Bubble team lift their support game.
Thanks @allenyang - this is the same answer we were given two weeks ago, but as constructed it doesn’t seem like the right tool for this specific use case (single point-in-time queries).
The pie charts are great to see what’s using capacity in your app over a period of time so you can try to squeeze out some efficiency gains or capacity reductions for your app as a whole.
Because of the way it’s set up though (starting date/time and no end date/time) getting data for a very small point in time event - like the spike-crashes that I’m seeing once a day or the spike in usage that @w.fly saw at 15:23 - requires hoping that the 10 minute window you’re viewing doesn’t dilute the underlying cause.
If you’re proposing that chart as a useful diagnostic tool for moment-in-time issues, I’d suggest adding a second date/time field that allows a user to say simply “I’d like to see what used capacity in my app between two points in time”.
Without that, the pie chart is still a useful tool, but for different purposes than you’re suggesting I/we use it for here.
This is very disappointing. As a non-coder myself, I’m wanting to learn Bubble in order to build my business on it, but if there are capacity issues and if the Bubble team aren’t immediately at hand to resolve the issues within days (it’s been a week now!), then it gives me and many others serious doubts about the feasibility of using and relying on Bubble whenever we do run into issues which will need specific support from the team. I’ll be keeping an eye on this daily to see if it gets sorted and for your sake I really hope it does because it seems you’ve done a lot of hard work on this platform and you’re only being let down by Bubble.
Thanks for the supportive words @Mm123 - I don’t necessarily view it as being let down by Bubble, but there’s definitely room for improvement.
At the end of the day, when you’re building any app there are going to be advantages and drawbacks to any stack you use. For non-coders like us, the major advantage of using Bubble is that we don’t need to source, hire and pay our own tech teams (dedicated or contracted) for every project we build out, which is conceptually the holy grail of app development and what makes the promise of Bubble so intriguing.
The drawback to that advantage though is that Bubble isn’t eliminating the need for a tech team, they’re consolidating it and owning that role internally by becoming the tech team of last resort for over 400,000 developers/teams.
Think of it like the DMV. You almost never need to actually show up at a DMV office for anything, but when you DO have to show up, you grab your number and you’re sitting there with half of humanity waiting to see someone for five minutes for something you’d have preferred to do on your own if you could.
So I fully understand we’re all making a tradeoff here and that we take the good with the bad. I’m sure this will even out over time. The team here is smart and the SignalFire guys are also about as good a VC as we could all hope to have behind Bubble. They’ll iron out the kinks, but at the moment there seems to be a growing chorus of voices saying (i) the number of “weird behavior” instances seems to be higher than it’s been historically, (ii) the transparency around what’s causing that hasn’t been great and (iii) the depth and level of support people are getting when they have real technical challenges they can’t fix themselves is less than they’d expect given the importance of the consolidated tech team function Bubble has taken on in all of our businesses.
From the perspective of what’s actually causing this specific issue in my app, now being on day 9 of self investigation and diagnosis with recurring crashes, I found one interesting piece of information on my own within a minute or two of yet another max-out / crash. I’m posting it here just in case anyone else sees a similar issue in their app.
My “Page Load” usage is disproportionally high due to high usage by an “undefined” page that is using as much or more capacity as our primary user interface. Our primary interface page is s2 and all other pages in our app had negligible usage over this same 10-minute time period, leaving the obvious question of what an “undefined” page in an app refers to. Perhaps the external libraries being called from a plugin? Hopefully we’ll find out.
