-
Notifications
You must be signed in to change notification settings - Fork 1.2k
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
Add ability to disable DiskManager #4330
Conversation
fcb49df
to
60b17a9
Compare
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.
Looks great -- thank you @tustvold
I really like that this should allow unified implementations of spilling operators as well as "error if we run out"
cc @crepererum
/// TempDirs to put temporary files in. | ||
/// | ||
/// If `Some(vec![])` a new OS specified temporary directory will be created | ||
/// If `None` an error will be returned |
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.
/// If `None` an error will be returned | |
/// If `None` an error will be returned (configured not to spill) |
let mut guard = self.local_dirs.lock(); | ||
let local_dirs = guard.as_mut().ok_or_else(|| { | ||
DataFusionError::ResourcesExhausted( | ||
"Cannot spill to temporary file as DiskManager is disabled".to_string(), |
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.
If a user sees this it may not be clear what is happening or what they can do about it.
What do you think about changing the message to something like
"Cannot spill to temporary file as DiskManager is disabled".to_string(), | |
"Memory Exhausted (DiskManager is disabled, cannot spill to temporary file)".to_string(), |
As a follow up (which I will do), I think it would be excellent to pass in a context string to create_tmp_file to make the error more specific.
pub fn create_tmp_file(&self, context: &str) -> Result<NamedTempFile> {
I will improve the error messages in a follow on PR |
Benchmark runs are scheduled for baseline = 07f65bf and contender = df8aa7a. df8aa7a is a master commit associated with this PR. Results will be available as each benchmark for each run completes. |
Follow on in #4348 |
Which issue does this PR close?
re #3941
Rationale for this change
I want to be able to enable a memory limit, but instead of spilling to disk return an error. The simplest way I could see to support this was to add the ability to create a DiskManager that is disabled.
What changes are included in this PR?
Are these changes tested?
Are there any user-facing changes?