The first thing I would do, before doing any work, would be to sit down with the client/your boss and explain just how bad the situation is, that drastic solutions are in order, and that it will take years to get this under control (with 1m LOC and 150 clients presumably all running customised software, this would take years to sort out even with a large team working on it). Unless they understand that from the beginning you will never get the backing you need to sort this out.
This might be one of the few occasions where a complete rewrite is justified (if you can keep scope limited to reproducing what you have). You've said the code is incredibly complex, and if the problem domain is incredibly complex too, you're probably stuck refactoring. If the problem domain is pretty simple (a CRM without too many extra features might be), you may be better starting with your smallest client who uses the product the least, asking for all the pain points, and things they love, about the current software, and writing a simple CRM to cover their needs which replicates the features of the current product, then gradually porting other clients over to the new system and adding new features to it, while keeping the old code-base in maintenance mode and fixing serious bugs only. If you do a rewrite you'd have to port the 150 clients over 1 by 1, and leave the other code in maintenance mode - your primary client may not be at all happy with that.
If that's not possible, you'll have to refactor it slowly while keeping the code in place, so the first step is to get it into version control, sort out a sane deployment strategy with testing servers, then try improving some small isolated areas of the code for one of the clients in isolation. Good luck!
Comments
The first thing I would do, before doing any work, would be to sit down with the client/your boss and explain just how bad the situation is, that drastic solutions are in order, and that it will take years to get this under control (with 1m LOC and 150 clients presumably all running customised software, this would take years to sort out even with a large team working on it). Unless they understand that from the beginning you will never get the backing you need to sort this out.
This might be one of the few occasions where a complete rewrite is justified (if you can keep scope limited to reproducing what you have). You've said the code is incredibly complex, and if the problem domain is incredibly complex too, you're probably stuck refactoring. If the problem domain is pretty simple (a CRM without too many extra features might be), you may be better starting with your smallest client who uses the product the least, asking for all the pain points, and things they love, about the current software, and writing a simple CRM to cover their needs which replicates the features of the current product, then gradually porting other clients over to the new system and adding new features to it, while keeping the old code-base in maintenance mode and fixing serious bugs only. If you do a rewrite you'd have to port the 150 clients over 1 by 1, and leave the other code in maintenance mode - your primary client may not be at all happy with that.
If that's not possible, you'll have to refactor it slowly while keeping the code in place, so the first step is to get it into version control, sort out a sane deployment strategy with testing servers, then try improving some small isolated areas of the code for one of the clients in isolation. Good luck!