We are having same issue. /tpa/api/consent is returning the following error:
“Something went wrong and we couldn't connect the app. Try again later.”
This has been happening for the last two days. Customers who are already logged in and connected are working fine. The issue only affects customers who need to log in or re-authenticate the app.
Has anyone found a solution/workaround?
mstornelli
That's the exact fault I hit on 4 September — same endpoint, same banner, same profile of existing connections working and new consents failing. It cleared after about a week with no change on my end. The fact you're seeing it again two weeks later means it's recurring.
To save you the fortnight I spent on it: it's not your redirect URI, scopes, or client. MYOB confirmed that to me in writing on 22 September — it's a timeout in their consent grant path under load. Their words: "We're continuing work on the underlying latency that can push consent grants past the gateway timeout under load." So they know, and it's not fixed.
Two things worth knowing. First, there's no workaround. The POST fails before MYOB issues the redirect, so there's no authorisation code to recover from anywhere. Second, your existing tokens are safe — MYOB also confirmed that a failed consent doesn't touch them, which matches what you're seeing.
What actually got engineering to look at mine: capture the failing POST in a browser with DevTools open, and post the x-amz-cf-id and timestamp (UTC) from the response headers here, tagging ROC_MYOB . That's the route that worked. It took a while, but it worked.
