[serve] Wait until replicas have finished recovering (with timeout) to broadcast LongPoll updates
#34675
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
edoakes
force-pushed
the
wait-to-broadcast-on-recovery
branch
from
34245dd
to
90b3730
Compare
Signed-off-by: Edward Oakes <[email protected]>
sihanwang41
approved these changes
Apr 25, 2023
| # Don't update http_state until after the done recovering event is set, | ||
| # otherwise we may start a new HTTP proxy but not broadcast it any | ||
| # info about available deployments & their replicas. | ||
| if self.http_state and self.done_recovering_event.is_set(): |
ProjectsByJackHe
pushed a commit
to ProjectsByJackHe/ray
that referenced
this pull request
May 4, 2023
…o broadcast `LongPoll` updates (ray-project#34675) When the controller recovers, all replicas are put into the `RECOVERING` state. These are not included in long poll updates for running replicas, which means we broadcast an update that effectively clears out all available replicas in all handles. This PR addresses this problem by avoiding broadcasting any updates until all replicas are fully recovered (or a 10s timeout is reached). We also wait to run the `http_state` update loop because if a new proxy is started, it won't be able to serve any traffic due to having no replicas available. Signed-off-by: Jack He <[email protected]>
architkulkarni
pushed a commit
to architkulkarni/ray
that referenced
this pull request
May 16, 2023
…o broadcast `LongPoll` updates (ray-project#34675) When the controller recovers, all replicas are put into the `RECOVERING` state. These are not included in long poll updates for running replicas, which means we broadcast an update that effectively clears out all available replicas in all handles. This PR addresses this problem by avoiding broadcasting any updates until all replicas are fully recovered (or a 10s timeout is reached). We also wait to run the `http_state` update loop because if a new proxy is started, it won't be able to serve any traffic due to having no replicas available.
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.
Why are these changes needed?
When the controller recovers, all replicas are put into the
RECOVERINGstate. These are not included in long poll updates for running replicas, which means we broadcast an update that effectively clears out all available replicas in all handles.This PR addresses this problem by avoiding broadcasting any updates until all replicas are fully recovered (or a 10s timeout is reached).
We also wait to run the
http_stateupdate loop because if a new proxy is started, it won't be able to serve any traffic due to having no replicas available.Related issue number
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.