The problem here is that given the current implementation of PI HA collective, ACE will need to connect to the primary server of the collective in order to start or stop calculations and hence you see the error as indicated by ACE scheduler and it does not proceed after that.
The reason for this limitation is that the ACE calculation status and information are stored in the Module Database (MDB) structure of the PI server. When a calculation is started or stopped the status of the calculation is updated in the MDB. Given that it is a PI collective, the changes are made on the primary PI server and then replicated to the secondary server(s). We can't make changes on MDB of a seconday server and hence we can end up in a situation like what you are facing.
To allow ACE to start working again quick, I believe the best thing to do would be to promote the secondary server to the primary server. There are some specific steps to do this and I believe our support engineers from techsupport team would be able to assist you on this.
Thanks for the explanation. It looks more clear now. But i was wondering that if this is the case then ACE will never work on HA ? till we make secondary servers as primary server.
Is there something OSI is coming up in future ? Apart from that , i saw that buffering is ON in the ACE server , so will all the calculation results (output tags) will automatically updated after connection ?
Sorry for the delay in reply. Technically ACE works with HA, but with a few things that should be noted. One of the important things to take note is what I've mentioned in the previous post.
Apart from that , i saw that buffering is ON in the ACE server , so will all the calculation results (output tags) will automatically updated after connection ?
We configure buffering on the ACE server, such that ACE can continue to calculate and buffering data to output tags even when one of the server is down. In your setup, the expected outcome should be that the calculation continues and result sent to output tag on secondary server while buffered for the primary server. ACE would not have to be restarted.
I guess the problem was that the buffer wasn't working as expected, which lead to ACE being restarted when the Primary server is still down. Consequently ACE cannot start up properly because Primary server is not available.
Is there something OSI is coming up in future ?
An AF Analytics scheduler is being proposed as part of a future release that would also enable users to configure analytics.
Potentially the calculations you are running today - depending on the calculation requirements - might be able to be built using configured analytics in the future!
May I ask what types of calculations you are performing in ACE?
For example arithmetic, roll-up, specific type of analysis