reload is difficult to use with instance jobs

Bug #1012938 reported by Tv
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
upstart
New
Undecided
Unassigned

Bug Description

If I have an instance job (such as ceph-osd), that all read a config file (such as /etc/ceph/ceph.conf) and I want to make them all reload their config (send SIGHUP), I can't just "initctl reload ceph-osd", I need to somehow obtain a list of the running instances, reload them one by one, and handle an edge case where they might have exited between me listing instances & running reload.

I could do an initctl list | mawk ... | ... etc, reconstruct the instance variables from the what's inside the parens, but that just seems too clumsy.

Related: https://bugs.launchpad.net/upstart/+bug/878322/

Revision history for this message
Clint Byrum (clint-fewbar) wrote :

Makes sense to me. The most elegant way would be to have a glob expression available like we have in the job syntax. Done right, '*' should mean "all instances"

initctl reload ceph-osd '*'

or

initctl reload network-interface 'IFACE=eth[012]'

Revision history for this message
Loic Dachary (dachary) wrote :

is it still undecided ?

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.