-
Notifications
You must be signed in to change notification settings - Fork 579
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
feat(meta): ensure each command is applied to exactly one database #19076
Merged
+468
−272
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
…bal-barrier-manager-context-trait
…bal-barrier-manager-context-trait
…bal-barrier-manager-context-trait
…bal-barrier-manager-context-trait
24 tasks
…-checkpoint-isolation
…bal-barrier-manager-context-trait
…bal-barrier-manager-context-trait
…g/separate-database-fragment
…g/separate-database-fragment
…-checkpoint-isolation-base
…g/separate-database-fragment
…-checkpoint-isolation-base
…single-database-command
wenym1
changed the base branch from
main
to
yiming/separate-database-fragment
October 29, 2024 03:37
hzxa21
approved these changes
Nov 1, 2024
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!
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
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Outdated suggestions cannot be applied.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.
I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.
What's changed and what's your intention?
Previously, we have a global streaming graph, and therefore the barrier command to the global barrier manager is applied to the whole streaming graph. However, as we are supporting isolation between multiple databases, if we have multiple databases sharing a same barrier command, the complexity will be greatly increased.
In this PR, we will change to ensure that each barrier command is applied to exactly one database. Currently we have the following command that may be applied to multiple databases according to its definition:
Pause
,Resume
, and command that contains aHashMap<FragmentId, _>
where theFragmentId
can be across multiple databases, includingRescheduleFragment
,SourceSplitAssignment
andThrottle
.For
Pause
andResume
, they are generated from either risectl or configuration change. For the ones generated from configuration change, we can associated the database id of the specific configuration change command to thePause
andResume
. For the ones generated from risectl, when handling these risectl command, we will list all active databases from catalog, and then sendPause
orResume
one by one to these active databases.For commands containing
HashMap<FragmentId, _>
, the strategy on the refactor will be to list all databases where theFragmentId
s belong to, and then divide them intoHashMap<DatabaseId, HashMap<FragmentId, _>>
, and then we apply the command to each database one by one.We have refactored the barrier command by the way. Previously, each barrier will be associated with a command. For normal barrier, we have
Command::Plain(None)
. In this PR, such variant ofCommand
is removed. Instead, we introduce an extraCommand::Flush
to handle the manually generatedflush
SQL. The database id of theCommand::Flush
will be the database of the current frontend client session. For periodic barrier, there won't be an associated command, and therefore each barrier will change to holdOption<Command>
instead ofCommand
, and it will carrySome(command)
when the barrier is generated by a scheduled command, andNone
when the barrier is periodic command.Checklist
./risedev check
(or alias,./risedev c
)Documentation
Release note
If this PR includes changes that directly affect users or other significant modifications relevant to the community, kindly draft a release note to provide a concise summary of these changes. Please prioritize highlighting the impact these changes will have on users.