Menno Lageman wrote: > Darren Reed wrote: >>> >>> config/enabled boolean kernel accounting state >>> >> >> In what situations would config/enabled be true and the SMF >> property (general/enabled) that reflects the state of the service >> be false, or vice versa, thus disagreeing about the state of >> accounting? >> >> Or put differently, why isn't this property redundant? >> > > Darren, > > the general/enabled property only determines whether or not the > configuration (whatever it may be) is restored upon boot. The > config/enabled property reflects the state of accounting in the kernel > (i.e. is the extended accounting subsystem active). > > It is perfectly valid to have a configuration with config/enabled > being false (e.g. after issuing 'acctadm -D task' to temporarily stop > writing accounting records while keeping the accounting file open) and > general/enabled being true. Upon boot this will restore the > configuration (set file name, tracked and untracked resources, do not > write records). So the property config/enabled is not redundant.
I must have picked the wrong property... there is one that reflects the current status...anyway... Why is acctadm the administrative interface here and not svcadm? Why wouldn't you use "svcadm disable -t system/acctadm:flow" ? Or is your point that "acctadm -D task" temporarily disables all three? And to ask the question I was asking in a different way, can config/enabled be different to the status of the service as observed using svcs? And if not, doesn't that then make config/enabled redundant? And if yes, does it make sense for the two to be able to disagree? Darren
