There is no control that ends the SSH hold on an earlier job in a serialized group and still lets the later jobs in that same group start.
Rerun job with SSH on any member reruns the entire serialized group and enables SSH on every member, not only the job you selected. Those jobs still run in order. CircleCI appends Wait for SSH after each one: about 10 minutes if you never connect, and up to 2 hours if you do.
Canceling that first job, or using Cancel workflow, marks the run canceled. Later jobs that requires a successful predecessor do not run.
This shows up hardest with job groups plus serial-group, including generated configs where isolating one job means changing the generator. Background: Debug with SSH.
Debug without stacking the wait
Do not SSH into the earlier members. After their steps finish they sit in Wait for SSH, then succeed, and the next member can start. Connecting to those jobs makes the wait much longer (up to two hours instead of about ten minutes).
If you only need the last job, rerun a pipeline that contains that job outside the serial group. For a generated config, that is a generator change, not a one-line edit in the UI.
Leave the earlier jobs untouched so the ten-minute hold expires on its own.
There is still no “stop SSH and continue” button. Ending the SSH session on an earlier job does not skip the rest of that job’s wait in a way that starts the next group member immediately.