Liane Praza wrote:
> How common is it during non-development scenarios that
> something ends up in maintenance and only an "svcadm clear; svcadm
> enable" is required to fix it?
I don't think that's the question.
I think the question is, given that the service failed, whether
administrators need to *explicitly* say that they've corrected the
problem (via svcadm clear) or if that can be assumed when they take some
other start-the-service action like "svcadm enable" or "svcadm start".
That is, how often will these scenarios occur:
1)
Service fails
Administrator fixes the problem
Administrator says "svcadm enable" or "svcadm start"
SMF refuses because you can't do that from maintenance
Administrator swears
Administrator says "svcadm clear"
Administrator says "svcadm enable" or "svcadm start"
2)
Service fails
Administrator says "duh, why isn't that running?"
Administrator says "svcadm enable" or "svcadm start"
Service tries to come up, fails catastrophically
3)
Service fails
Administrator says "duh, why isn't that running?"
Administrator says "svcadm enable" or "svcadm start"
SMF refuses because you can't do that from maintenance
Administrator blindly says "svcadm clear; svcadm enable"
Service tries to come up, fails catastrophically
Personally, I'd say that the manual intervention of requiring a new
"enable" is sufficient. I've seen far too many people blindly click
through warning messages to believe that adding a rote step will
significantly improve the situation. Administrators prone to ignoring
the real problem will quickly learn that you always do "clear" before
"enable". (In fact, I've seen product wrapper scripts that do this.)
I would *not* consider this to be part of the "start" discussion; if
anything I'd say that it applies *more* to "enable".
