Feature Request: Option to limit the RAM usage during the rescan of Scale-out Backup Repositories

[enigma361]

Hello Veeam R & D,

I'd like to suggest the addition of a feature for limiting RAM usage during Scale-out Backup Repository rescans. In our environment we have SOBRs for storing SAP HANA plugin backup files and the rescan of these consumes an enormous amount of RAM which sometimes causes the backup server to crash. Obviously this is a really bad scenario since all our backup and restore jobs crash too when this happens.

Therefore we'd like to see this limiter feature to limit RAM usage during rescans to e.g. 30 % of the available RAM or any other amount we or other users chose. In any case the performance of running backup / restore jobs shouldn't be subordinated to rescan jobs.

A good addition would also be to warn the user that the repository rescan can potentially crash the backup server if the repository is very large or has millions of files (as can be the case with SAP HANA backup files) and thre are many jobs already running. It unfortunately happend a few times already that someone from our team started a rescan process and the server crashed as a consequence.

We already had opened a case for the problem but it was found that the product works as designed at this stage (Veeam 10a). There is just too much going on on the server at times and our SOBRs have too many files to rescan them while many jobs are running.

Please consider our request and change the product to include a limiter and also show a warning that rescans can consume a large amount of RAM and are potentially dangerous.

Thank you very much.

Best regards,
Nils

[Andreas Neufert]

Hi Nils, thanks for sharing. I asked one of the team members to comment on this
but would like to ask you share here the support ticket number for our reference.

How many task slots are configured for the Scale out Backup Repository extends? Maybe set to unlimited?
How much RAM is there?

I think it would be better to reduce the RAM usage than working on limiting the processes on the other end.

[PetrM]

Hello,

Most likely the issue during rescan of SOBR with SAP HANA backups occurs due to full read of backup metadata. This logic will be optimized in one of our future releases: in order to decrease RAM consumption only metadata header will be used, it is enough for rescan purposes. Therefore, there will be no need for additional limiters.

I'd prefer to let our support team to provide all necessary technical details and to estimate the possibility to get the private fix for this issue or to find an appropriate workaround.
Please don't forget to paste the support case ID over here so that we can make sure that it's on the right track.

Thanks!

[enigma361]

Hello Andreas, hello Petr,

thank you for your replies. That sounds good. I have opened a new support case regarding the possible private fix or workaround. It is: # 04382328. The old case I mentioned was # 04382328. I also mentioned the latter in the description of the new case.

Thank you.