Merge takes forever (surprise) and job runs way longer than "Job Interval" period. Merge Operation Questions

[BackItUp2020]

I've been STRUGGLING to get good copies off-site to our cloud provider. The hang up at this point seems to be the merge operation. I have a two-part question...

1) If a merge job doesn't complete, I assume the merge operation has to start over again at the next cycle since new data will likely be copied. Correct?
1a) Does this mean if a merge doesn't complete after many, many cycles, it is likely the merge will take significantly longer as time goes by because the changed data *could* be much higher?
2) I have a copy job that is scheduled to run every 2 days and it has run now for 5 days (131 hours) and is at 4% on the merge operation. Why has it not stopped an retried after the 2 day cycle?

It has been so frustrating with these copy jobs. And the hint of an internet issue, maintenance on the cloud storage side, or a large change on the source side and BAM, things get logjammed, and what happens more than not is a full reseed is needed. Its impossible to keep up with.

Would a replication job help limit these issues at all?

Thanks for any input!

[PetrM]

Hello,

1) - Correct. Merge happens after every job run to apply retention policy and make room for the most recent restore point.
1a) - Basically, merge duration depends on the amount of data but I don't think the merge will take significantly longer if Fast Clone is used on the target. For instance, Fast Clone is available for REFS and XFS.

2) It is by design but I think it would be best to ask your cloud provider to open a support case and to investigate the root cause of slow merge with our engineers.

I guess the replication job might be a good workaround because it allows us to follow the 3-2-1 rule as well. You can have a look at the remote replica from backup to reduce the load level on the production VMs which will be 2x higher if 2 jobs process the same VMs.

Thanks!

[BackItUp2020]

Thank you for the insight!

I have a few cases open on slow copies and merges currently. As far as the merges, the latest thought is that I need to schedule periodic maintenance because of my use of forever incrementals. So, (as expected), I have changed that setting and kicked off an active full for my problematic jobs. Its going to take some time (weeks) to finish though. My jobs have been hovering around 2MB/s w/ "Target" bottleneck so we'll see how it pans out.

The real bummer is that I have had pretty much no reliable data in our cloud provider for months.