I could be wrong, but based on my own experience, when a project starts off this poorly, it was not ready to go live and the fix usually takes much longer than a few weeks.
The typical pattern is to get into fire fighting mode and start creating bandaids to get past critical problems, but that consumes your time leaving little time to fix the actual problem. As you get past the first wave of problems, the users begin using different parts of the system creating new issues.
I’ve seen things like this drag out for many months or even a year. If there is a fundamental design flaw, portions of the system could just be turned off and handled manually/semi-automatically behind the scenes.