Pbm restore high ram usage

Hello! i have 3 node replica set and i have 1 full backup

pbm status –mongodb-uri=
  - prod-nosik/ pbm-agent v1.6.0 OK
  - prod-nosik/ pbm-agent v1.6.0 OK
  - prod-nosik/ pbm-agent v1.6.0 OK

PITR incremental backup:
Status [OFF]

Currently running:

FS  /opt/mongodb-backup
    2021-12-29T07:54:26Z 404.91GB [complete: 2021-12-29T08:36:59]

and i’ve create another Replica set for test restores, then i launch pbm restore and restore fails because of OOM, i tried to set 4gb limit, but it doesn’t help
my vm specs:
CPU: 4
RAM: 8
Disk: 1tb /var/lib/mongo

1 Like

Since you are doing the restore, you will not need much WT cache (set 1G for WT cache) and also you can alter the pattern of eviction too. This should help with the restoration process and you can revert the changes after restore. However, I see less impact from WT cache in front of OOM issue as you set its value already + MongoDB uses extra memory for other internal processes and connections etc from filesystem cache. You will need to make sure you keep the DB server dedicated to MongoDB. If not, it is better to check if any other process utilizes more memory in the server. Pls check your dmesg output for more details.
For more details, refer this blog - https://www.percona.com/blog/2020/05/05/tuning-mongodb-for-bulk-loads/

1 Like