I have one particular volume for which we don’t have a snapshot schedule on. The volume snapmirrors to another site, therefore the only snapshot that should be present is the one generated by the snapmirror job. However, it seems to want to keep hold of them, therefore eventually the volume fills and goes offline. I have 4 other volumes configured in the same way, that do not display this behaviour. The only difference is that the rate of change of this particular volume is much higher on a daily basis because it is a backup drive
I have set autodelete to delete the oldest first but this doesnt seem to delete. The volume is set to auto grow/shrink
It sounds like the rate of change on this volume is causing the snapshot, that's created by snapmirror, to fill up the volume. Having autodelete enabled for the snapshot will not help here as that snapmirror snapshot is locked by the system. Have you thought about changing the snapmirror schedule on that volume to replicate the data more frequently? The other option would be to turn off snapshot deletion so that the volume only autogrows.
Let's talk about the way snapmirror replciation works.
When you configure snapmirror and perform the initial replication on a volume a snapshot occurs. That snapshot then grows as changes occur on the volume. Once you update the snapmirror relationship a new snapshot occurs, data is transfered and the first snapshot is deleted.
It sounds like the snapmirror replication isn't completing, because the source volume is filling and goes offline. If snapmirror was able to replicate all of the data then you would only end up with a single snapshot.
Have you tried to modify the schedule so the snapmirror replication occurs more frequently?