Solved

Tape job issues if job can't complete in time



Show first post

32 comments

Userlevel 7
Badge +17

Veeam has to make sure that the complete backup chain is written to tape. You cannot restore from the incrementals when then full from last Friday is missing.

So, you have to let the job finish. And yes, this can cause a missed incremental or an incremental at another tome of the day.

Do you have one tape drive only? If you have more, you can try to configure the tape job to use multiple drive in parallel to get the data faster to that tapes.

Userlevel 4

So what happens if it misses a FULL backup, as happened here? It will just try the next scheduled execution time, even though it will never finish the job before the next backup job starts? I have backup jobs that can get kinda large (like 7-10 TB), and no way would they finish before the next scheduled weekday execution time, if they don’t finish on the weekend. How do you get around that?

 Block storage is the better option here, but it depends on what you have available.  Part of the bottleneck is the DD unfortunately and the way Veeam synthesizes to tape.

 

The DD is what I have for storage. Been using one for like 12 years here, first with EMC Networker and now with Veeam (as we are transitioning). Even with other storage, writing 10TB to tape in 1 day is probably not really feasible, unless maybe you had multiple LTO-8 drives and 10G fiber ethernet (maybe) ..

Userlevel 7
Badge +20

So what happens if it misses a FULL backup, as happened here? It will just try the next scheduled execution time, even though it will never finish the job before the next backup job starts? I have backup jobs that can get kinda large (like 7-10 TB), and no way would they finish before the next scheduled weekday execution time, if they don’t finish on the weekend. How do you get around that?

There is no way around that unfortunately as that is how the tape service runs.  One thing around it would be to have another backup location as going to DD for primary is not a best practice and reading from that to tape is what can take much longer time.  Block storage is the better option here, but it depends on what you have available.  Part of the bottleneck is the DD unfortunately and the way Veeam synthesizes to tape.

Userlevel 4

So what happens if it misses a FULL backup, as happened here? It will just try the next scheduled execution time, even though it will never finish the job before the next backup job starts? I have backup jobs that can get kinda large (like 7-10 TB), and no way would they finish before the next scheduled weekday execution time, if they don’t finish on the weekend. How do you get around that?

Userlevel 7
Badge +20

How do you have the tape job configured?  GFS?  From what I know Veeam does this to ensure that a full backup chain makes it to the tape, and you don’t have a broken one.  So, this is why Veeam is processing the full backup to get that to tape and will then proceed to the incrementals if you have those checked in the tape policy.

No GFS (actually, I’ve never used a “real” GFS scheme in 30 years of doing backups … always done just saving EOM and EOY tape backups, all others recalled and overwritten after 60 days).

Actually, due to legal constraints, at the moment we do not overwrite *any* removable media. So no old backups are ever recalled from offsite and overwritten and re-used. Just sent offsite ...

So there’s no way to accomplish this? With our old program, what got written to tape was whatever backup was performed in the specific tape job time period (i.e., within the last 24 hours). If I wanted an earlier backup (that might have been missed) sent to tape, I had to set a specific job that wrote that specific saveset (media set).

This is just how the Veeam to Tape jobs work.  There is nothing you can control on what writes out to tape per day/schedule.

Userlevel 4

How do you have the tape job configured?  GFS?  From what I know Veeam does this to ensure that a full backup chain makes it to the tape, and you don’t have a broken one.  So, this is why Veeam is processing the full backup to get that to tape and will then proceed to the incrementals if you have those checked in the tape policy.

No GFS (actually, I’ve never used a “real” GFS scheme in 30 years of doing backups … always done just saving EOM and EOY tape backups, all others recalled and overwritten after 60 days).

Actually, due to legal constraints, at the moment we do not overwrite *any* removable media. So no old backups are ever recalled from offsite and overwritten and re-used. Just sent offsite ...

So there’s no way to accomplish this? With our old program, what got written to tape was whatever backup was performed in the specific tape job time period (i.e., within the last 24 hours). If I wanted an earlier backup (that might have been missed) sent to tape, I had to set a specific job that wrote that specific saveset (media set).

Userlevel 7
Badge +20

How do you have the tape job configured?  GFS?  From what I know Veeam does this to ensure that a full backup chain makes it to the tape, and you don’t have a broken one.  So, this is why Veeam is processing the full backup to get that to tape and will then proceed to the incrementals if you have those checked in the tape policy.

Comment