Urgent: bad request (in production, prevents users from purchasing our product)

JavaScript
8

Thanks, Allesandra. This is exactly what we need. We’re looking into it.

Regards,
Mark

1 Like
9

I’m sure you know already but right now it is all down (so also the app is impacted, users cannot create accounts, etc.)

10

We were making changes to get the business logic to work. Could you please check and let me know if you’re experiencing any problems? I tried the GET request in SubscritionNoCoreData and it times out, I assume because there is no logged in user. Please let me know.

Regards,
Mark

1 Like
11

thank you Mark, I was able also to go through with the payment operation. Can you let me know more about what happened (in case it happens again, then I can provide better feedback).

I appreciate your time and prompt resolution

12

I am glad it works now. The root cause was the migration to a faster file system we did earlier today. Everything went smoothly except for the coderunners access to the new file system and that’s where the hickup was when model.json could not be found (while the file was visible in console).

I apologize for the inconvenience this has caused, but we’re in better shape now.

Regards,
Mark

1 Like
13

@mark-piller and team, this is happening again now

not sure if something changed after addressing this issue: Password reset not working (backendless 3, happening every few months) (cc @sergey.kuk )

but we are getting a lot of users that are unable to purchase the product (I just tried myself now and got the same error, bad request for the purchaseorder object, which has an oncreate, so most likely again a business logic issue)

Thank you for your help

Please somebody acknowledge seeing this

14

Hi Alessandra,

Do you happen to have a sample request that is failing? Preferably with the error it returns?

Mark

1 Like
15

thanks Mark, in the developer console, under Network in Chrome, where I see the error, I have the following: ““code”:0,“message”:“Model Descriptor not found (/opt/backendless/repo/9088af67-1648-1726-ff09-a51862bf1a00/files/servercode/JS/v1/model.json)”}”

16

Could you please go through requests in the Network tab and find one that is failing?

17

that’s what I just did, showing that message (it seems the same issue as the last time if you scroll up in this thread)

18

Could you please right click the request and select “export as curl” ?

19

Thanks. We are looking into it

1 Like
20

any updates? thank you

21

Not yet. It is taking some time. Will post an update as soon as I have one

1 Like
22

We made a few changes. Could you please check on your side?

23

it seems that all business logic calls now return this:

{“code”:15000,“message”:“Custom business logic execution has been terminated because it did not complete in permitted time - 20 seconds”}

24

Does it happen with all business logic calls? Even the basic ones?

25

we have only this purchaseorder oncreate because it must work to call stripe and get people to pay, no other business logic is currently in place due to its unreliability (everything is disabled to make sure that at least who has paid, can access what they paid for)

the only other task is for admin purposes, a find on users, that also fails with the exact same error now

how did this stop working when it was all good 2 days ago?

26

The primary reason is your app is running on an ancient version of Backendless. The app continues to grow, both in database and in file storage. As it happens it becomes significantly harder for us to maintain it.

You need to seriously consider upgrading to the latest version as incidents like this will continue to happen.

Back to the problem… what is the typical invocation time of the logic that times out now?

Regards,
Mark

27

Mark, it’s a quick call that does not do anything but calling stripe, and it happens on our smallest table (purchaseorder)

there is some new problem here that doesn’t have anything to do with how long it’s supposed to take.

The other call on users simply counts how many registered in the past 24 hours, week and month, and the users table is also much smaller than tables that are working very well with a lot more data (but without business logic)

I get your point on the older version, but we discussed this already, how can we upgrade if we cannot even use 2 mobile SDKs at the same time for the two versions? There needs to be a phase in which we have both systems running, otherwise this would be a complete nightmare