Yep, that's it. Should listen to you more... :) I think we need that backported than, right?
> Am 15.02.2018 um 14:11 schrieb Yann Ylavic <[email protected]>: > > On Thu, Feb 15, 2018 at 2:03 PM, Stefan Eissing > <[email protected]> wrote: >> On my MacOS: >> >> r1824180 of mod_slotmem_shm.c exhibits the problem described >> r1824179 of mod_slotmem_shm.c works >> >> This is really strange since r1824180 in 2.4.x is identical trunk, afaict. >> >> Tried restarting server, manually deleting the slotmem files, rebooting >> the machine, all the same. >> >> Anyone a guess where I should look? > > Did you try applying r1667707 ? > Your previous log showed empty workers ("AH01172: > balancer://https-local: worker () rewritten to /002.jpg"), which looks > symptomatic to me. > > With new mod_slotmem_shm it's really crucial to not have duplicates, > previously maybe the attach worked around this (not debugged there > though, only a guess).
