About a month ago we upgraded from 7.19 (or the latest version of 7) of Commerce all the way to 9.19.
Since then we've noticed that catalog/product information changes are not reflected until the app pools are recycled on IIS. Typically this would take 15-20 minutes on 7.15.
My gut tells me it's something with our Caching configuration, we use the standard out of the box caching on episerver. Is there anything other than ecf.catalog.config that is used to govern caching now?
Any information would be helpful, this is a strange one.
our ecf.catalog.config (everything is set to 15 minutes)
What UI/API do you use to update the catalog data and in what UI/API are the changes not reflected? Different representations of catalog data (like DTO and Content) and different applications (load balanced servers, frontend vs commerce manager) are synchronized on update so changes should be immediately visible independent of timeout settings. The latter part is dependent on working remote events, e.g. working UDP communication between local servers (other mechanisms used in cloud).
http://world.episerver.com/documentation/Items/Developers-Guide/EPiServer-CMS/75/Event-management/WCF-event-management/ check wcf event settings.
Hello Magnus and K Khan,
We have numerous methods, we already checked the events and did notice that the event settings were wrong after our upgrade and believe we have that sorted.
No matter what method we use to update the product information, it's not reflected immediately. Most recent test was making the change in the commerce UI on one server and testing on that same server and the values didn't change until we refresh the app pools.
It has us a little stumped. I'll look at the links posted, thank you.
So it is the rendered page templates that are not upated? Are you using any custom models etc that you cache in your implementation? Are you using output caching or a reverse proxy (Varnish etc)?
Are you using Amazon sqs event queues? If so in Amazon.events 9.0 ( I think) there was a bug where it deletes all the sqs queues. This means the cache invalidation message doesn't get sent to the other servers. The latest version fixes it.