I guess on sending side they will get timeout errors. so they could build some mechanism from sending side ? is there any best practice around it whenther to handle at sender or in PI ?
I guess on sending side they will get timeout errors. so they could build some mechanism from sending side ? is there any best practice around it whenther to handle at sender or in PI ?
If SAP system is down, then it can be just a system re-start in which case the system will be back in 10-15 minutes OR it can be an outage where it might be down for hours.
So, you can't handle all the situations by increasing the time out either in PI or in the sending system. I would say that you should keep the time out as recommended for sync cases. If messages fail in PI, better to re-start from the sending system.
so someone needs to keep an eye on PI regarding sync messages if they dont get a response back from synchornous message and sending side needs to send it again.
so someone needs to keep an eye on PI regarding sync messages
Just configure alerts, so that you don't have to manually monitor the system. Then a direct alert mail can be sent to sender mentioning that resend of message is required.
1. Is it a planned downtime? If yes, then it is always better to inform the portal users well ahead before you go for a downtime. Synchronous messages cannot be restarted in XI/PI.
2. If it is an unplanned downtime then you must resend the whole data from the portal, keeping in mind that you are not using any BPM at anywhere in the design. Otherwise, I wholly agree with Prateek that you configure alert for unplanned downtimes and also if possible , use BPM so that you can restart the particular workitem at any point of time.
Though I have a doubt on how will you be able to restart a BPM workitem at which there is a synchronous call to ECC since synchronous calls are not restartable in XI/PI.
Depending upon the volume of payload and volume of messages, there can be a other way(although it is not a solution and must be avoided unless you have no other way but to use it) to work this out. You try to push the payload from RWB->Component Monitoring->Integration Engine->Test, but remember, this is not a solution at all and this should not be made a practice. It involves effort from the XI guy, this is not automated and also if the payload is heavy, then you will not be able to work this out. Also, there is a high possibility of data tampering, even unintentionally.
This is as good as pushing the data from the portal and the response will not go back to any portal session. So your endusers won't get any practical benefit out of this, only the data those failed during the down time can be posted to the ECC.
Just thinking about alerts : alerts will be sent to a predefined set of technical team.
The users of website will be across the globe ,so alert message might not reach each and every user. Instead i was just wondering if PI could send some fault message that SAP is under scheduled maintenance ,this message could go directly to the end user.
although its just a thought and not sure about it.
You can also configure alerts or generate specific emails whenever SAP is down. it really depends on how your team wants to handle application systems downtime.
Incorrect (if it is about resending from PI). You cannot resend/restart synchronous message in PI. It has to be resent from original sender. However, if you have a BPM, the restart of sync step is possible.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.