So, if I understood correctly, both are the way to determine which bean to autowire if there are multiple candidates. So what exactly is the difference?
Read @Primary as the "default".
If a bean has @Autowired
without any @Qualifier
, and multiple beans of the type exist, the candidate bean marked @Primary
will be chosen, i.e. it is the default selection when no other information is available, i.e. when @Qualifier
is missing.
A good use case is that initially you only had one bean of the type, so none of the code used @Qualifier
. When you then add another bean, you then also add @Qualifier
to both the old and the new bean, so any @Autowired
can choose which one it wants. By also adding @Primary
to the old original bean, you don't have to add @Qualifier
to all the existing @Autowired
. They are "grandfathered" in, so to speak.
@Primary
is also good if e.g. 95% of @Autowired
wants a particular bean. That way, only the @Autowired
that wants the other bean(s) need to specify @Qualifier
. That way, you have primary beans that all autowired wants, and @Qualifier
is only used to request an "alternate" bean.