Azure is more restrictive than Jira when it comes to linking between work items, thus it’s not always possible to synchronize some links from Jira to Azure. We synchronize links in a sorted manner (starting with those having smaller ID).
This behavior means that when two links cannot exist with each other on the same work item in Azure we are only going to synchronize one with a lower id.
Example
Opposite links in Jira are allowed to point to the same work item from one work item (for instance in default Jira configuration, A can be blocked by B and A can blocks B).
Azure does not allow such behavior for opposite links thus only the link with the lower id is synchronized.