Partition space checking

[pj888888888888]

So I have gotten burned a few times with there not being enough space on the partition that LVM creates the snapshots on, and the agent more or less "freaks out" and just continues to run, and will not allow itself to be cancelled/closed (not sure if there is a way around that)
Would it be possible to add a feature where the backup job checks for available space before running?
Perhaps even checking what the settings are in the ini file and if the space is there, the job runs, if not, it fails and alerts you that there isnt enough space for the job to run?

I think that would be pretty neato if that were a feature/function in this agent.
(Along with cluster support too of course)

[PTide]

Hi,

So I have gotten burned a few times with there not being enough space on the partition that LVM creates the snapshots on

What do you mean by that? Veeam does not use LVM snapshots, do you create them manually? I guess it is as simple as you having an impression that the Agent uses LVM snapshots, but it never hurts to double-check

<...>and the agent more or less "freaks out" and just continues to run, and will not allow itself to be cancelled/closed (not sure if there is a way around that)

Does the agent do that in the beginning of the job, or in the middle? There are two possible situations:

- agents is unable to place an initial portion of a snapshot
- agents is unable to place a subsequent portion

What error do you see in the session log?

Thanks

[pj888888888888]

I probably misspoke with the LVM comment, disregard that, I jotted this up rather quickly so it wouldn't be another thing that slipped my mind.

Its actually right in the middle of the job that it seizes.
What I have been able to deduct is that it has to do with the partition space.

The unfortunate thing is that there isn't an error in the session log, I have redirected most of the jobs to not use the root partition being when this happens, it has caused certain systems to crash.
The job more or less runs out of control and is unable to be killed, in the logs, it just looks like its still trying, but no activity is happening.
The processes are still very much active when you look in top however.

I had an aha moment when i looked at the space on the partition i directed the agent to use in the ini file.

(I also put snapshot space constraints in also just to make sure I could control the potential blowouts....which is what I am calling this issue )

Im just curious if it would be possible to put it into the job to do a space check against the INI file for max size and check the partition beforehand.....might be a headache saver if possible.

[PTide]

Hi,

The unfortunate thing is that there isn't an error in the session log

So, in other words, there is not one bit of an error or a warning that the partition has run out of space or that the snapshot has been overflown or cannot be allocated. It does not look as an expected behaviour and I would really appreciate if you opened a case with a support team and posted your case ID.

Regarding snapshot constraints - those only work with a "common" snapshot type. In most cases "stretch" snapshot is used instead, unless you use some old OS, or enforce it manually via .ini file. What is your distro version btw?

Regarding built-in checks - it makes sense and should be possible to implement for 'stretch' snapshot too. Just let us investigate why didn't the job give you any errors in the session log.

Thanks

[pj888888888888]

Actually, I had issues with stretch snapshots, and was advised by support to minimize pain points to use common snapshots.
I had a few jobs that run away, and caused the system in question to reboot due to those, so I opted to take the safer road and use the common snapshots and implement the max snapshot size in the ini to stop that from happening again.

Although....it appears if the space isn't there, the job runs away regardless.

Kind of seems like its an either/or decision on this.

But it would be nice to have that kind of warning/safeguard in place if its possible......or a good idea in the near future.

[PTide]

Your request is clear and makes sense. I believe that such space check must have been implemented already, but I'll double-check with the team and will put your feature request across.

Thanks!