> > We have a scheduler that gets populated with all provisioning tasks and > the necessary parameter values for the API calls, in order, and with > dependencies. >
Couple of suggestions. First, make sure the provisioning calls do a done check, for those instances where manual intervention takes place, the api can return an all OK, the scheduler can continue on it's own without having to manually mark the task as complete. Second, have a means to update the tasks with relevant provisioning data, in the advent you need to update the parameters being passed to a call. Like ooops, it's trying to setup passing IP 10.0.0.1, but that's in use somewhere else and failing, so the task can be updated with a new IP. Third, have a means to mark the task as done manually, for those instances when all goes blah. For long running operations, you may want to use a transaction based API call. Call API, jobs starts, API returns transaction ID, go back and check operation intermittently. On Tue, May 28, 2013 at 8:49 PM, henry ho <henr...@doteasy.com> wrote: > > To automate provisioning of services, a server must make a series of API > calls. > > From time to time, one or more servers would not respond, and it will > result in services being partially provisioned. We would then manually > figure out which part has been provisioned and carry out the remaining > provision steps manually. > > What strategy would you use to mitigate this problem and improve the > robustness of the system? > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~| Order the Adobe Coldfusion Anthology now! http://www.amazon.com/Adobe-Coldfusion-Anthology/dp/1430272155/?tag=houseoffusion Archive: http://www.houseoffusion.com/groups/cf-talk/message.cfm/messageid:355852 Subscription: http://www.houseoffusion.com/groups/cf-talk/subscribe.cfm Unsubscribe: http://www.houseoffusion.com/groups/cf-talk/unsubscribe.cfm
