sync request is done in 3 steps

1. register cmhandle to cps-core and put entry into fragment table.
2. fetch module ref from dmi plugin
3. fetch unknown module schema from dmi plugin

Initial sync might fail due to various reasons e.g.

• node is down
• dmi-plugin in down
• network partitioning etc

There must be some way to resync the cmhandle irrespective of previous failures.

Current problem is if sync request is sent again to resync the node if usually find the

entry in cps-core/fragment table due to step 1 in above para and return entry already found exception in logs.

Team consensus needed to fix this problem, possible solutions are

Mandatory step: currently we dont have any flag to know that cmhandle is registered or not , create a flag in dmi-registration.yang and set it true after all

3 steps are success.

Then

1. read above flag if false bypass (already existing cmhandle in fragment ) exception in sync request and execute other 2 steps. OR
2. provide some other rest end point to resync. (this could be covered by planned functionality for https://jira.onap.org/browse/CPS-507)