On 9/1/2011 8:20 PM, Angus Salkeld wrote: > On Thu, Sep 01, 2011 at 06:33:25PM -0400, Robert Knight wrote: >> As I have reported in the Wiki, there are troubles with the SysV init >> scripts on Fedora 15. The defect is caused by the change in F15 to a >> tmpfs for /var/run. Thus, the target directory for the pid's does not >> survive a reboot. So, some way is needed to (re-)create /var/run/glance >> and /var/run/nova or the startup scripts need to be changed to just put >> the pid's directly into /var/run, which will be there. Does anyone know >> whether there is a standard resolution to this? > Can't the init scripts just "mkdir -p /var/run/nova" before starting the > binary? That was indeed the solution I adopted. What I was asking was whether there was a conventional solution -- almost all of the other packages that are already present do not build their own subdirectories in /var/run. >> The Short-Description and Description fields in the SysV init script for >> openstack-glance-registry (as well as the other comments) is incorrect. >> >> The systemd packaging guidelines say that the systemd unit files should >> be sent back upstream. Does anyone know whether upstream would accept this? > I converted the init scripts to systemd units in f16, I haven't > submitted these upstream as they don't seem to have any init scripts > in their repo's (other than for xenserver). I'm glad to hear it. In which src.rpm are they? > -Angus > >> _______________________________________________ >> cloud mailing list >> cloud@xxxxxxxxxxxxxxxxxxxxxxx >> https://admin.fedoraproject.org/mailman/listinfo/cloud > _______________________________________________ > cloud mailing list > cloud@xxxxxxxxxxxxxxxxxxxxxxx > https://admin.fedoraproject.org/mailman/listinfo/cloud Regards, Robert _______________________________________________ cloud mailing list cloud@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/cloud
