how bug is the aggregate? do you see any messages that would indicate problems? if the write workload is very high it can take a long time until it switches back to sync. offlining the plex can end in the resync starting over again. Can you try to see if you can reduce write workload?