We have multiple groups and it seems its alphabetically determining what to show as the primary group in the user info command. For example, I have myself with parent groups admin and default, and according to the yaml file my primary group is default, yet in user info it shows admin. If I do user parent switchprimarygroup to default, it (correctly) says that I already have that set as the primary group. If I remove the admin group as a parent, then the user info command correctly shows default as primary group.

Then, I tested further to prove this is some alphabetical sorting related bug of some kind: I added admin back as a parent but then I also created an additional group called aaatest and added it as a parent group, and indeed, the user info command then shows aaatest as the primary group (instead of admin or default) while the yaml file still has it set to default. I did the same with a group called test and because its lower in the alphabet it did not show as the primary group in the user info command.

So all in all it seems to only be a visual/textual output bug in the user info command and hopefully nothing more than that.

Versions in use are: Spigot 1.15.1 and LuckPerms 5.0.50

Have you given the groups weight?

https://github.com/lucko/LuckPerms/wiki/Command-Usage:-Group#lp-group-group-setweight-weight

A higher weight means the group is more important than an inherited group of lower weight

No, I did not give the groups any weight.

give them weights and then try

Is there any update to this issue

Thank you for your suggestion. I added weights to the groups and that indeed "fixed" the issue.

However, upon inspecting my config.yml it turns out primary-group-calculation: parents-by-weight was set instead of primary-group-calculation: stored. I don't remember having set this, maybe it was the default at some point in time. Setting it to stored fixed the issue without needing to give weights to the groups.

primary-group-calculation: parents-by-weight has been the default for over 2 years.

Oh, well the first paragraph here made me think it was different at some point (e.g. determined automatically based on server type detected at run time).

Oh. Wow. Thanks for the find. That is mightly outdated!