-
Notifications
You must be signed in to change notification settings - Fork 1.9k
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
[Remote State] Upload each metadata attributes file instead of single global metadata file #12468
Labels
Comments
shiv0408
added
enhancement
Enhancement or improvement to existing feature or request
untriaged
labels
Feb 26, 2024
8 tasks
github-project-automation
bot
moved this from 🆕 New
to ✅ Done
in Cluster Manager Project Board
Feb 28, 2024
github-project-automation
bot
moved this from ✅ Done
to 🏗 In progress
in Cluster Manager Project Board
Mar 13, 2024
@shiv0408 thanks for elaborating on the issue. Please share some numbers of improvement as well. thanks! |
Thanks for reopening the issue @sarthakaggarwal97. I have added the benchmark results in linked PR. We see around 50-70% improvement in writing Incremental Metadata, see PR #12190 for more details. |
github-project-automation
bot
moved this from 👀 In review
to ✅ Done
in Cluster Manager Project Board
May 16, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
Is your feature request related to a problem? Please describe
Currently, we are storing our remote cluster state in a global metadata file and index metadata files for each index and maintain all the info in manifest file. As the cluster grows and cluster state size grows with more usage, the global metadata size also increases.
Whenever we trigger a cluster state update, we need to write the updated metadata on remote. If the global metadata file size has increased and the incoming change is also in global metadata file, we will upload the whole file again, for a small change of settings as well, thus increasing the cluster state update latency.
Describe the solution you'd like
We propose that we split the global metadata file into following following components:
This way if only a setting is modified, we don't update other files. If multiple files need to updated, those are updated in parallel, which is again better than upload full file.
This change will help in majorly decreasing cluster state update latency in big clusters.
Related component
Cluster Manager
Describe alternatives you've considered
No response
Additional context
No response
The text was updated successfully, but these errors were encountered: