Memory issues at the weekend

[lando_uk]

Hi

Our VBR server that is also a proxy and repository crashed with some memory errors this weekend. It's the first time this has happened, nothing has changed from other weekends, but this one is a end of month where we do some copy jobs health checks what I enabled a couple of months ago due to disk issues.

This physical server has 40GB RAM and a 30GB Pagefile. On the docs, if the server has all 3 roles, it says it needs 10GB for each concurrent task. so it looks like the health checks along with the normal backups broke it. Does it need more RAM or will just increasing the pagefile sort this out?

Thanks
Mark

[foggy]

Mark, how many jobs/tasks were running at the moment the server crashed? Indeed, you need to make sure the server has sufficient amount of RAM to accommodate all the roles and be able to serve each task assigned to it.

[lando_uk]

Normally only 2 concurrent backup jobs runs, but with the copy jobs doing their health checks which take all weekend, there may have been 8 jobs running.

[foggy]

You can spread the health checks for different jobs over different weekends to avoid them from being triggered at once at the end of month.

[lando_uk]

I thought I did, but I shall revisit the settings to make sure.

I have some spare dimms for these servers, so its no big deal. Looking at the memory usage on Veeam One, its always topped out at the weekends for the last few months, maybe I'll just use whatever spares I have until it never goes to 100% again.

Thanks

[foggy]

You can also limit the number of concurrent tasks for repository server to control the load.

[lando_uk]

Just an update. I upgraded our VBR server from 40GB to 70GB and it has shaved a few hours from the weekend synthetic full's, According to VeeamOne, server memory now peaks at about 60GB during the busiest period.