Automatic HugePage configuration is available starting with FlashGrid Storage Fabric version 19.02. After resizing VM, HugePages will adjust accordingly if this feature is enabled. HugePages are configured only on DB nodes (not configured on quorum or storage nodes).
If you enabled automatic configuration of HugePages in FlashGrid Launcher during the cluster/server deployment, no further action is needed.
Note: Prior to version 19.02 of FlashGrid Storage Fabric, this feature was not available and HugePages must be configured manually by editing /etc/sysctl.conf file.
To enable automatic HugePages configuration:
- Add or edit the following parameters under
[memory]section in/etc/flashgrid.cfgon the first DB node:
[memory] configure_hugepages = True total_db_memory_pct = 80 total_sga_max_size_pct_of_db_memory = 60
Please note that HugePages is configured as a product of total_db_memory_pct * total_sga_max_size_pct_of_db_memory. Based on the above values, it is 80% * 60% = 48% of total memory. Check recommendations for memory sizing
- Deploy the updated configuration:
# flashgrid-cluster deploy-config
- Update the following parameters at the database level:
SGA_MAX_SIZE = <size of HugePages> SGA_TARGET = <size of HugePages> USE_LARGE_PAGES = only
Note: If you have multiple databases sharing the same cluster, then the sum of the SGA_MAX_SIZE or SGA_TARGET (whichever is greater) parameters across all databases must not exceed the amount of memory allocated for Huge Pages.
4. Reboot one node at a time. See instructions for a graceful reboot here.