Howard Chu wrote:
Howard Chu wrote:
The consumer seems to be OK in general - it ignores the NEW_COOKIE if it's equal to or older than its current ctxcsn. However, I've now seen another case where a consumer receives the NEW_COOKIE message from one provider before it has received the real update from any providers. So again, when the real update arrives, it is ignored.
Appears to be the ITS#5973 patch, not 5972.
Reverting the syncrepl part of the ITS#5973 patch appears to let test050 pass for me now. But looking at the logs, there are a lot of NEW_COOKIE messages being sent (and there ought to be none in this scenario). Also, I'm seeing a lot of entries propagated without any CSN in the sync cookie. The results look correct, but there's still something flaky here.