The rules can include "on failure" actions, I suppose. Probably the
local system should be able to override that.
Markus Hofmann <hofmann(_at_)bell-labs(_dot_)com> 08/10/01 03:01PM >>>
while we've been talking about OPES failure handling in the past, none
of these discussions are reflected in our current OPES drafts - it
became clear in London that we've to change that. We need to identify
possible failure cases and include clear mechanisms in the OPES
framework for dealing with such failures!
First question - which draft(s) should include a discussion on failure
handling? Failure cases certainly are related to "OPES Use Cases and
Deployment Scenarios". On the other hand, failures also relate to the
failure of specific OPES components, which are introduced in the "A
Model for Open Pluggable Edge Services" draft. Comments?
Second - consideration of failure cases also has impact on the policy
requirements. For example, it seems that rules need to include
instructions for how to behave in certain failure cases. More
specific, how should an OPES device behave when a callout server
returns an error? Is this something that should be specified by the
rule author, as he/she probably knows best whether the original (not
serviced) content is acceptable for delivery or not. Comments?