Skip to Content
avatar image
Former Member

Alerts and retrigger mechanism in HCI

Hi Experts,

We have developed a scenario in HCI where the integration is between Ariba and an on premise PI system .

We have configured alerts using exception subprocess (referring to other blogs posted here about alert configuration), where we have used a mail adapter to send error alerts to our mail ids.But what we noticed is , the exception subprocess gets called only for certain errors - like faulty script used in iflow,data faullt encountered within the iflow etc. But for connection issues (eg. if target system is down, channel not responding , temporary connection issues),the message fails and gets logged as a failed message in the message processing logs with no call being made to the exception subprocess. Please clarify what errors invoke the exception subprocess and if alerts can be configured for other errors like connection issues with sending/receiving systems.

Also, please clarify if a retry/retrigger mechanism can be configured for such connection errors(In PI we can resend the message/retry attempts can be configured at channel level). Is there a similar functionality available in HCI.

Thanks and Regards,


Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

1 Answer

  • Best Answer
    Mar 20, 2018 at 01:16 AM

    Hi Tina - Unlike PI there is no option to reprocess the failed messages in CPI.

    Regarding the Alerts if you have solution Manager probably you can leverage that to get the alerts for the Connectivity issues etc.,

    Other option to handle the error might be to catch explicitly with in your iFlow like using Request Response adapter where the exception will be thrown.

    Add comment
    10|10000 characters needed characters exceeded

    • Former Member

      Hi Rajesh,

      Thank you for the information. So, for any failures in HCI due to connectivity issues, messages would have to be resent from the source system?

      Yes, we did try the request-reply mechanism for alerts and that worked. But our scenario is actually an asynchronous one. Would using request-reply have an impact on this.