Hi, I'm having a similar problem with my experiment at https://mili2nd.eu/xj3b.
It's also set to sequential subject id and group id numbering, but when I sent it out to people for testing, it skipped a lot of subject ids. I thought that the /groupassignment attribute might have something to do with this, but setting this to subjectnumber didn't appear to work.
I read that if you open the link to the experiment, but don't actually start it, Inquisit assigns you a subject id anyway, even though it doesn't actually show up in the logs. However, my subject ids were numbered 3, 5 7, 9, 10, 11, 15, 16, 23, of which only 9, 10 and 11 I tested on my own computer. I also checked with the people who tested it for me and if they clicked the link more than once, it wasn't enough to explain this pattern.
Do I need to also reupload my experiment to fix this?