-
Notifications
You must be signed in to change notification settings - Fork 15
adaptivemm: initial vm.vfs_cache_pressure tuning algorithm #49
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
adaptivemm: initial vm.vfs_cache_pressure tuning algorithm #49
Conversation
drakenclimber
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The logic looks solid to me. I had a few minor comments but no showstoppers.
056d6ac to
418a8a3
Compare
|
I think I've addressed all your comments from the first review |
418a8a3 to
1d65f3e
Compare
Implement a simple vm.vfs_cache_pressure tuning algorithm. The algorithm works by using a user supplied hint which is the PREFER_OBJECT_CACHING option in the configuration file. This hint indicates if the user would rather have their system preserve dentry and inode object caches at the cost of higher memory utilizaiton. Some workloads are strongly impacted by the level on dentry and inode caching so it may benefit these workloads to have the mm system deprioritize reclaiming these caches. If this option is selected, the inode and dentry numbers are monitored and if they are projected to increase (via the slope of a best line fit), the sysctl is lowered by 10% and if they are projected to decrease, the sysctl is increased by 10%. The opposite behavior occurs if PREFER_OBJECT_CACHING is specified to 0. Signed-off-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
1d65f3e to
1a17ed7
Compare
drakenclimber
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
Reviewed-by: Tom Hromatka <tom.hromatka@oracle.com>
Implement a simple vm.vfs_cache_pressure tuning algorithm. The algorithm works by using a user supplied hint which is the PREFER_OBJECT_CACHING option in the configuration file. This hint indicates if the user would rather have their system preserve dentry and inode object caches at the cost of higher memory utilizaiton. Some workloads are strongly impacted by the level on dentry and inode caching so it may benefit these workloads to have the mm system deprioritize reclaiming these caches. If this option is selected, the inode and dentry numbers are monitored and if they are projected to increase (via the slope of a best line fit), the sysctl is lowered by 10% and if they are projected to decrease, the sysctl is increased by 10%. The opposite behavior occurs if PREFER_OBJECT_CACHING is specified to 0.