Backup to tape, linked job delay, and most recent chain

[bdoe]

I think I'm looking at this right, but I wanted to check before I start changing a bunch of schedules.

I have several large file servers, mostly around 20TB each. The backup jobs are reverse incremental, and have backup to tape jobs on an LTO-6 autoloader. Each job does an active full once a month on Saturday (which is a 14 hour job), and incrementals runs through the week (not the weekend). The tape job usually takes 40+ hours. To fit it all in one tape job, up until now I've been doing (for example) the FS-01 job on one weekend, maintaining 7 days of rollbacks, and scheduling the tape job for the following weekend, also Saturday. This didn't work great since sometimes there would be two .vbk files and the autoloader would try to write 40TB to tape, without enough tapes.

So, you can imagine I was excited about the new feature in v9 that only writes the most recent backup chain to tape. I'm thinking there's a better approach to all this, hinging on that feature. I'm thinking I could move the active full jobs to various Fridays, and schedule the linked tape job for the same time. Then, in the schedule options, set the tape job to wait for up to say 1200 minutes - 20 hours will be more than enough. That way, the tape job can automatically start sometime Saturday morning, and have plenty of time to finish over the weekend instead of wrapping up Monday like they're prone to now. This will also mean Veeam won't waste time and fail jobs because it tried to write 40TB+ of .vbk to a pool that doesn't have that kind of space. And as an added bonus, the tape will be the most recent .vbk, not one that was written and added to over the course of a week.

It makes sense in my head. Am I reading this right, or would this layout not work?

[veremin]

The described scenario looks valid. Don't forget to enable "process latest full backup chain only" in the advanced setting of backups to tape job and you're good to go. Thanks.

[bdoe]

I'm trying to implement this, but the schedule page for the tape job won't let me set the job to wait longer than 777 minutes. That seems like a random number, is there a reason for this or is it a bug?

[veremin]

Looks like a GUI limitation, which me may address down the road. For now you can schedule a tape job to start later that source backup job does instead of starting both at the same time and use maximum value (777) as a waiting interval. Thanks.

[bdoe]

Hmm. Is there a proper way to get it submitted as a bug? If it's just a GUI issue I like to think it'd be easy to fix - easier said than done, I'm sure!

The problem is going to be the full backups take around 14 hours, sometimes more. 777 minutes gives me just under 13 hours. I don't want them all running at once, so they start around 7pm and are spread through the evening, about 90 minutes apart. Sometimes incrementals are in the terabyte range and take over an hour, so it has to be spread out. If I set the tape to start at 11:59pm, it may work for some, but it seems like the ones that start later may run out of time. I can probably make it work but it'd be a lot simpler if I could just tell the tape job to wait as long as needed.

Thanks!

[veremin]

Is there a proper way to get it submitted as a bug?

You've already done it. Your request for increasing the said value has been counted.

The problem is going to be the full backups take around 14 hours, sometimes more. 777 minutes gives me just under 13 hours. I don't want them all running at once, so they start around 7pm and are spread through the evening, about 90 minutes apart. Sometimes incrementals are in the terabyte range and take over an hour, so it has to be spread out. If I set the tape to start at 11:59pm, it may work for some, but it seems like the ones that start later may run out of time. I can probably make it work but it'd be a lot simpler if I could just tell the tape job to wait as long as needed.

You might either make a tape job start later than 11:59 pm to increase its chances of archiving a job running longer than expected or use "as new backup files appear" and prohibit its execution on all days but Saturday. In the latter case, the job will copy point as soon as it appears. So, once the first source backup job creates restore point, the backup to tape job will automatically archive to tape, and then wait till the second source backup job finishes its activity and so on and so forth.

Thanks.

[bdoe]

I may have missed something in the setup, but this doesn't appear to be working as I'd hoped. The active full backups are set to the intended days, and the associated tape jobs are scheduled for 11:59pm on those days and set to wait 777 minutes. The tape jobs are also set to process only the latest full chain. However, each time, the tape job does run while the active full is happening, but lasts under a minute, backs up nothing, and reports success. This has happened for three weeks now, so either I made the same mistake on each job or there's some other problem. Any ideas?

[veremin]

I'd recommend opening a ticket and letting our engineer confirm whether everything is setup properly. Thanks.

[bdoe]

I had a feeling that'd be the next step, thanks. Case is 01706217.