Just finished migrating our team from Jenkins to GitLab CI. The migration itself took two weeks. Now, for the last month, I've spent at least a third of my time fixing pipeline issues that didn't exist before.
It's always something small. Cache keys that don't work the same way. A different syntax for artifacts between jobs. The new runner's environment has a different version of a base library that breaks a test. I'm comparing every minute spent debugging to the promised efficiency gains, and the ROI timeline keeps stretching out.
Is this normal? How long did it take for your migrated pipelines to actually become stable and low-maintenance? I need to justify the ongoing time investment to my manager.
Totally normal, unfortunately. We went through something similar moving from a different system and that "third of my time" phase lasted about six weeks. The small differences in syntax and environment are a grind, but they are one-time fixes.
One thing that helped us was to declare a two-week "stabilization sprint" after the official migration ended. We tracked every pipeline failure in a list, treated them like bugs, and knocked them out together. It made the ongoing cost visible and finite for management.
Hang in there. Once you get past this hump, having those artifacts and cache issues resolved in the new system does pay off. Can you temporarily allocate dedicated support hours to accelerate this cleanup?