SyncIQ run Policy when Snapshot of Source Dir is taken

Question

Hi

We are planning on implementing SyncIQ policy execution based on "Whenver a snapshot of the source directory is taken", and this got me thinking:

If I have a repeated snapshot schedule for, lets say every 5, 10 or 15 minutes, with an expiration set to 2 hours. What happens if, for some reason, SyncIQ misses the sync (maybe the replication target is offline) when it's supposed to? What happens if the last sync is based on the snapshot that's about to expire next?

Does a snapshot with a set expiration actually really expire if the system's SyncIQ policies are based on this snapshot-chain or is there a mechanism in place that prevents expiration and automatic delete of (non SIQ type) snapshots that are still the base for SyncIQ compare+replication?

This does not really get covered in Dell Isilon/PowerScale documentation...

Thanks for any input!

Peter_Sero · Accepted Answer

A snapshot can get one ore more 'locks' from other parts in OneFS, which means deletion is deferred beyond expiration time until all locks are released. You can see whether a given snapshot has locks:isi820-1# isi snapshot view s16 ID: 16 Name: s16 Path: /ifs/data Has Locks: No ... So just try it out (the OneFS simulator will do); a regular snapshot should get a lock once the triggered SyncIQ job kicks in.makes sense?-- Peter

Peter_Sero · Answer

Sounds like you might want to consider 'archiving SnapshotIQ snapshots with SyncIQ', whicht is covered in section 9 of this whitepaper: https://www.delltechnologies.com/resources/en-us/asset/white-papers/products/storage/h8224_replication_isilon_synciq_wp.pdf  hth -- Peter

CendresMetaux · Answer

Hi Peter

I read your referenced document already and unfortunately this does not exactly answer my question.

My question is more of a technical one than an implementation oriented one.

Does Isilon retain a SnapshotIQ snapshot that is set to expire or expired already, if it is the base for the last successful SyncIQ policy run (which is set to run whenever a snapshot of the source directory is taken)? Or, would such SyncIQ policy error out (resync needed) if it fails to (example if target cluster is not reachable for an extended period) run again before the last "base" snapshot's expiration time is reached and therefore got deleted?

CendresMetaux · Answer

Hi Peter

Now thats the Answer I was looking for! I hoped to see something like that, a locking system or something.

In this case, it is worth to try and I will proceed and create a test environment and hopefully take this to production

Thanks a lot!

CendresMetaux · Answer

Just to add information:

Besides CLI, also in the WebUI each SnapshotIQ based snapshot that is currently used/referenced by SyncIQ for replication, gets visually and explicitly marked as locked. In the snapshot details it is pointed out that the snapshot is locked by SyncIQ (SIQ-sched-man-lock). See following screenshot:

Isilon

SyncIQ run Policy when Snapshot of Source Dir is taken

Was this post helpful?