On a system with huge_page enabled it seems more reasonable to just mount the huge page config into the container, that will avoid any issues that we can faced when trying to deactivate huge_page which doesn't seems to fully work anyway.
For example for the Postgres container with we disabled huge_page on the Postgres conf the container start but resource shared_buffer 400Kb and max_connections 20 seems limited,
fixing permissions on existing directory /var/lib/postgresql/data/pgdata ... ok
creating subdirectories ... ok
selecting dynamic shared memory implementation ... posix
selecting default max_connections ... 20
selecting default shared_buffers ... 400kB
selecting default time zone ... Etc/UTC
creating configuration files ... ok
running bootstrap script ... ok
performing post-bootstrap initialization ... ok
When mounting the hugepage into the container we can see the shared_buffer and the max connections are ok
fixing permissions on existing directory /var/lib/postgresql/data/pgdata ..ok
creating subdirectories ... ok
selecting dynamic shared memory implementation ... posix
selecting default max_connections ... 100
selecting default shared_buffers ... 128MB
selecting default time zone ... Etc/UTC
creating configuration files ... ok
running bootstrap script ... ok
performing post-bootstrap initialization ... ok
syncing data to disk ... ok
To reproduce
1- Create a EKS cluster
2- SSH to one of the node and enable hugepagee
On a system with huge_page enabled it seems more reasonable to just mount the huge page config into the container, that will avoid any issues that we can faced when trying to deactivate huge_page which doesn't seems to fully work anyway.
For example for the Postgres container with we disabled huge_page on the Postgres conf the container start but resource shared_buffer 400Kb and max_connections 20 seems limited,
When mounting the hugepage into the container we can see the shared_buffer and the max connections are ok
To reproduce
1- Create a EKS cluster 2- SSH to one of the node and enable hugepagee
3- Deploy yba to the specific host with node selector
helm upgrade --install yw-test yugabytedb/yugaware --version 2.17.2 -n yb-platform --values your_yba-value.yml
4- if you don't mount the hugepage you will see that YBA can be deployed because of max connections reached.