On Tue, May 1, 2018 at 11:15 PM, Jim Jagielski <j...@jagunet.com> wrote: > Considering that we have some regressions in .33 which > will soon be fixed (these are the 2 noted ShowStoppers)
The fix for PR 62308 is being tested and we should be good soon, I think. I don't think PR 62277 is a regression/showstopper, the OP reports an issue on slotmems after an upgrade from 2.2 to 2.4 (slotmems didn't exist in 2.2, and the mod_proxy_lb quite changed in between). Also, it looks more like an IPC-SysV "limitation" (on some Solaris version?) than a bug in our code, maybe should we consider a move to POSIX sems by default in trunk/2.next? By the way, I modified the tool (attached in bz) to show what I suspect are spurious collisions in ftok()+semget() with the OP's balancers' names, but he switched to POSIX sems so I'm not sure he will test any further. Possibly Rainer could try to run it on his Solaris(es) to confirm or not (no specific version mentioned in bz...). > as well as a limited number of "other" changes to the > codebase, maybe now is a Good Time to consider a 2.4.34...? +1, once these are addressed. > > I offer to RM using Daniel's updated release scripts so > we can get a different set of eyes using the scripts. Thanks! Regards, Yann.