cantor.2 at osu.edu
Mon Apr 11 12:09:44 EDT 2016
> > Not establishing a consistent username after authentication can cause a lot
> of problems, so that's hardly just a workaround, but that said, this seems like
> a bug in the storage service.
> Ok, thanks I'll file a JIRA issue for this.
To be clear, my point is that if it "worked", it should be creating a separate set of consent records for a username that differs from another. If that's not case sensitive, perhaps the problem is with the database schema, but either way, something seems off here.
The correct configuration in any event is certainly to normalize the name, because you wouldn't want the user's data entry affecting which records were used. That part isn't a bug, it's a mistake in configuration.
> For me c3p0 worked much better than hikari, hikari pool was "dead" in
> less than 24h. c3p0 works for at least few weeks at a time.
> I'll test with dbcp.
No experience with hikari, I stuck with dbcp 1.2 because pooling that works is pretty much in the "don't touch it again" space. I don't even know if dbcp 2 works.
More information about the users