I am wondering if anyone else experiences this problem, or have found a way
to deal with it...

We are using Jackrabbit Oak version 1.4.10, and running two HTTP
server-fronted instances in Prod. Both instances are configured to talk to
the same Mongo database. HTTP servers are load balanced.

The problems is when I create a node using instance 1, and after that
quickly access that same node using instance 2, instance 2 reports the node
as non-existent. It usually takes a second or two for the instance 2 to
"see" the added node.

This is very bad news for us, as create-and-immediately-retrieve pattern is
used a lot in our application. We currently found a dirty workaround by
running DocumentNodewStore.runBackgroundOperations() on every new Session -
but this does seem like an intended way to work in a cluster.

Any thoughts or suggestions appreciated.

-- 
Thanks,
Eugene

Reply via email to