Is there any possibility to intentionally persist data to workspace from a ssh job? (by default it is disabled)
Warning: skipping this step: SSH is enabled, not persisting data to the workspace.
For example, we have a long pipeline for release builds and our first step obtains credentials and saves them into a workspace. Then in later jobs these credentials are taken and used
If a mistake happens during the first step (for ex. some credentials are outdated) we’ll have to rerun the whole pipeline from the very beginning to correct the failed job.
If we had a possibility to update workspace files manually the time and credits would be saved.
You should open up a support ticket for this question as it will need an internal staff member with knowledge of the system internals to comment.
If the main issue is credential management have you looked at the idea of storing the values outside of the build jobs, so they do not need to be passed between jobs within the workspace? This would allow you to modify the value on the fly while a job is just reporting an issue with the key that it has retrieved - it can then just reload the current value and try again within a loop, rather than you having to ssh in.
I use a third-party solution called Doppler as my secret store and key/value store and it does not take much work to integrate it into a job.