When synchronising work logs from Atlassian Jira, ChronoLink no longer seems to randomly ignore certain issues.
It turned out that when an issue has more than 100 worklogs, it cannot be found reliably with search criteria involving the worklogAuthor or worklogDate fields. The reason is a limitation that Atlassian had to put in to ensure good search performance.
To fix this, we have switched to another approach that no longer relies on those fields. As a side effect, this new approach also improves performance.