Implement worker autoscaling #12
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
bengl
suggested changes
Jul 16, 2019
talbenari1
force-pushed
the
talbenari1/autoscale
branch
2 times, most recently
from
d2ce306
to
a135eef
Compare
talbenari1
force-pushed
the
talbenari1/autoscale
branch
2 times, most recently
from
d3452c7
to
5d77fa2
Compare
Rather than running each worker in a single separate thread, I'm now creating a new threadpool for each worker. Since threadpools will only spawn new threads as needed, we don't have to create our own load pressure detection method. One downside of this is that the first request to each worker is slower, so to deal with that (and to fix issues with workers that fail to start) I spawn an empty task onto the isolate's threadpool to wake up a single isolate instance. Note that the while the threadpool will always be created, only one instance will be created by default (i.e. autoscaling will be disabled) until we figure out a good story for applications that rely on global state. For now, the feature will be hidden behind a '--experimental-autoscale' flag.
talbenari1
force-pushed
the
talbenari1/autoscale
branch
from
5d77fa2
to
e66eb75
Compare
bengl
reviewed
Aug 5, 2019
| @@ -37,6 +37,7 @@ pub fn handle_inbound((req, tx): Message, origin: &str) -> impl Future<Item = () | |||
|
|
|||
| req.into_body() | |||
| .for_each(move |chunk| { | |||
| let context = get_context(); | |||
There was a problem hiding this comment.
What do you need these in this file for?
There was a problem hiding this comment.
Since the body chunks are received asynchronously, there is a possibility that the context will have been destroyed before the request finishes processing, which can cause segfaults. These will panic with sane error messages instead.
bengl
approved these changes
Aug 5, 2019
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Rather than running each worker in a single separate thread, I'm now creating a new threadpool for each worker. Since threadpools will only spawn new threads as needed, we don't have to create our own load pressure detection method. One downside of this is that the first request to each worker is slower, so to deal with that (and to fix issues with workers that fail to start) I spawn an empty task onto the isolate's threadpool to wake up a single isolate instance.
This should speed things up, like, a lot. Especially in cases where most of the time is spent in userland.
A couple of other minor things: