We have a single-node FAS 8020 that I need to take down tonight for maintenance (installing a new HBA) - I have 2 HA clusters that snapshot and mirror ~125 volumes to this replication target FAS periodically throughout the day and night.
Are there going to be any problems with taking the replication target FAS host down for a couple hours, or will the snapshot replication and mirroring jobs just sort of queue up while the target is down, and then start chugging again afterward?
Sorry, I couldn't find anything in the docs that covers this aspect, and I'm hardly a NetApp wiz.
No, they do not queue up. If transfers are in progress, they are simply aborted.
Therefore, the best practice is to: 1) Quiesce the replications if the transfers are not in progress when the "maintenance starts" and resume them once the maintenance is over. This simply puts the relationships in suspended state.
2) However, If the transfers are in progress, they will be aborted when the filer goes down. It's up to you to either abort them manually before the maintenance start or have them aborted when the filer loses contact.
Just quiesce them all from the destination filer: Ones that says "Quiesced", sorted. Ones that says "Quiescing", just abort them or they will be aborted.
From the doc: The snapmirror quiesce command disables future transfers for a SnapMirror relationship. If there is no transfer in progress, the relationship becomes "Quiesced". If there is a transfer in progress, it is not affected, and the relationship becomes "Quiescing" until the transfer completes. If the current transfer aborts, it will be treated like a future transfer.