Calculating Digests

[bhagen]

I have some replication jobs that have been working just fine. This morning before they ran, I edited each job and changed the text in the "Description" box.

Now every job is "Calculating Digests" and it's going to take "forever" to complete...several days, anyway, rather than 15-20 minutes.

Is this seriously supposed to happen after such a trivial change to the job?? What can I do to stop this now that it's started?

[fortune_KN]

Hi bhagen,
Normally, this change should not cause calculating digest to happen.
Have you not change other parameters like source of the replication for instance?

[bhagen]

Changing the description of the job is the only change that was made.

[bhagen]

Arggg!! The "Repository for replica metadata" got changed as well!

That solves it.

Apologies for the confusion!

[fortune_KN]

That it is!
You are welcome! You just have to wait now.

[bhagen]

No way; I'm not waiting a week for these to finish...I'm stopping them "gracefully", and then I'll copy the replica data folder to the repository I pointed the jobs at and re-start the jobs.

(I know; "gracefully" has never worked for me either...meaning, it usually takes as long as the job would have taken in the first place. So I normally give it a try, and then just kill the job "immediately".)

[fortune_KN]

Even by doing that way, you will not avoid the digest calculation as you are still using the same job.
Another way could be to create a new job and use mapping feature to leverage existing Replica.

[foggy]

Generally, copying replica metadata should help to avoid digests re-calculation, the job should pick it up. However, I'm not sure about the terminated job run, it might cause digests re-calculation as well.

[bdufour]

i think if you terminate the job, on the next run it will throw the cant find dirty block info in the previous restore point - the job wont use cbt.

[bhagen]

Thanks for your help so far, everybody!

I've copied the replicas folder to the new repository.

I've confirmed that terminating jobs caused digests to re-calculate. I've also confirmed that if I edit one of those terminated jobs and check the "replica seeding" box, then map replicas using "detect", the job runs with no "dirty block" issue nor digest re-calculation. So that's good.

Now my question is: if I edit a job and uncheck the "replica seeding" box, will it calculate digests again?

I guess the better question would be: is there now any need to uncheck the "replica seeding" box?

[bhagen]

...the job runs with no "dirty block" issue...

Actually, this turns out to not be true. I've started the next job that I'd terminated, which I'd edited by checking the replica seeding box, and it indeed "Cannot find dirty block information in the previous restore point, CBT will not be used".

I guess that makes sense since the previous restore point didn't finish.

However, it isn't calculating digests! Good thing too; this is a 15TB VM.

[bdufour]

good deal, glad this worked out for you.

blake

[bhagen]

However, it isn't calculating digests! Good thing too; this is a 15TB VM.

Sheesh; this isn't true either! It just started calculating digests.

So now I'm going to have a gap in replication while the entire 15TB vm is re-calculated (this'll take a couple days). Or I could delete the replica from disk and start a new job, but that would delete all the restore points.

I don't understand the mechanics of why this happens, but I sure wish it wouldn't, or there was a better way to make changes to replication jobs without basically starting the job over from scratch!

[bdufour]

dont delete the replica, that would be an even worse situation as you would have to re replicate the entire vm.

[fortune_KN]

Sheesh; this isn't true either! It just started calculating digests.

So now I'm going to have a gap in replication while the entire 15TB vm is re-calculated (this'll take a couple days). Or I could delete the replica from disk and start a new job, but that would delete all the restore points.

I don't understand the mechanics of why this happens, but I sure wish it wouldn't, or there was a better way to make changes to replication jobs without basically starting the job over from scratch!

Hi Bhagen,
Digest calculating occur everytime an action which can change VM or Disk ID/size is carried out because veeam need to compare the "virtual" new VM and the target replica to keep the integrity of replica chain.
Running an active full replication will make it avoid Digest to be re-calculated. To do so, you don't need to delete all your replica as you can use them for Seeding/mapping.

Must of the time, running a new active full replication in such situation is better, but it can depend on the infrastructure in place (WAN link, vNetwork, size of vDisk and VM,...)

[bhagen]

Running an active full replication will make it avoid Digest to be re-calculated.

I don't see a way to run an active full replication job?