A note about mod_shib2 and mod_http2

Cantor, Scott cantor.2 at osu.edu
Thu Mar 8 17:31:54 EST 2018

> From a further reading of some of our internal communication (six months is a
> long time ago) it seems the CPU load last time was probably related to a known
> httpd issue which we patched.  Then we discovered the SP problem, for which I
> am now back to thinking the existing patch works.

I would think so, it was a very obvious deadlock once it was pointed out. I think there's a bit of a race possible that I need to revisit, but it isn't one that should hurt much unless the reload delay on a piece of metadata was so short that it tried to fire off a reload before it fetches the first copy. We need to start the refresh thread after the initial fetch.

> I'll keep chasing down this rat-hole and see if I catch anything interesting but if
> you don't hear back from me, assume the xmltooling patch does indeed work.


-- Scott

More information about the users mailing list