We use version flexible snapmirror for all SnapMirror jobs, and we use a custom policy that includes the "all_source_snapshots" label. I recently set up two new volumes on a 9.3 system and replicated them to a 9.2 system. They worked fine until I needed to rename the source and destination volumes. After that change, the snapmirror jobs fail when run on a schedule (they work most of the time when I run them manually the next day). The errors complains that the snapmirror snapshot already exists on the destination. Even weirder, I just noticed that not all of the regular snapshots are being sent. There are 7 snapshots retained on both volumes, but only 5 of them show up for one of them and 6 for the other. Any ideas?
Thanks for replying Paul. The snapshots are very small, only MB in size. What makes it weirder is the missing snapshots are in the middle ... for example snapshots 0 and 1 are replicated, 2 is skipped, then 3 - 6 appear. Replication on one of the volumes failed last night. When I ran it this morning manually it was fine.