Hello Team,
We have a system that runs Hybris 6.6, recently upgraded to v1808, it uses a clustered environment with Jgroups. Every week, we do a data load to our system that uses the following logic,
load the data into the inactive catalog
Once the load is complete, switch the inactive catalog to the active one.(Previous active catalog will now change to the new inactive catalog)
Cleanup the old values in the current inactive catalog
All this activities are done in the backoffice node with the store being up in another node. Soon after we switch the catalog and start the cleanup, we get exceptions in the store logs such that the data is missing and the loaded data is not shown in the storefront. While checking the data from backoffice, we can see that the data is indeed present in the active catalog.
The exceptions lasted for around 2 hours and then it recovered automatically and started displaying the proper data.(We couldn't try clearing the caches in the nodes as we did not suspect this initially)
The data is fetched using the following steps. When user visits the store, and navigates to the respective product, query(flexible query) the item using the current active catalog version for the items loaded as per the aforementioned steps.
To check if this is a cache issue, we did the same steps of switching the catalog (no data load involved) in a similar environment with clusters. And as soon as the catalog was switched we ran the following groovy script
catalogService = spring.getBean("catalogVersionService")
println catalogService.getActiveCatalogVersionForCatalogId("xxxxxx").version
from the HACs of all the cluster nodes individually and found that the catalog version had been switched successfully.
Can this be an issue with the cluster cache not getting invalidated and containing the flexible query caches for the items?
Bluesky
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.