8000 Questions to answer after the VE varying size tests · Issue #903 · cometbft/cometbft · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

Questions to answer after the VE varying size tests #903

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.

Sign up for GitHub

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

Jump to bottom
Closed
4 of 6 tasks
lasarojc opened this issue Jun 1, 2023 · 3 comments
Closed
4 of 6 tasks

Questions to answer after the VE varying size tests #903

lasarojc opened this issue Jun 1, 2023 · 3 comments
Assignees
@lasarojc
Labels
chore documentation Improvements or additions to documentation

Comments

@lasarojc
Copy link
Contributor
lasarojc commented Jun 1, 2023 •
edited
Loading

A TODO for PR #888
Consider in the #578

  • Memory usage seems to indicate a trend up.
    - Run baseline for for longer to clarify.
  • During the 5 minute interval to clean up the mempool, some nodes actually have the mempool expanding.
    - Hypothesis: these are seed nodes, which frequently have to catch up. We need to investigate further; possibly filter messages in seed nodes.
  • If the mempool size drops quickly, why is the memory slow to empty? Slow GC?
    - The load created by the load generator seems to be a small part of the actual load.
  • Runs with larger extensions use many rounds.
    - Check the effect of running with longer timeouts.
  • The app causes proposals to be large, not only VE.
    - Run with the app modified so the proposals are not increased?
  • what is the effect on block sizes?
    • blocks are of size 2 x Extension Size x Quorum size.
@lasarojc lasarojc added documentation Improvements or additions to documentation chore labels Jun 1, 2023
@lasarojc lasarojc self-assigned this Jun 1, 2023
@lasarojc
Copy link
Contributor Author
lasarojc commented Jun 1, 2023

These images show the memory usage over 5 twenty minute runs.
They don't indicate any up trend in memory consumption.

min image

max image

avg
image

actual values
image

@lasarojc
Copy link
Contributor Author
lasarojc commented Jun 2, 2023

We have run with longer timeouts and deltas (6s) and still we see some high rounds.

image

For more details, we looked into maximum and minimums round numbers.
The maximum shows that pretty much in every height, some node required more than one round to decide.

image

And the minimum showed some node decided on the first round in the vast majority of rounds.

image

The average does not add much information

image

But the 50% quantile shows that not decided on the first round is not an isolated problem.

image

Seed nodes are not the culprit here.

image

The final graph shows the bandwidth usage, showing that proposals are using a lot of bandwidth

image

@lasarojc lasarojc mentioned this issue May 3, 2023
Closed
31 tasks
@melekes melekes mentioned this issue Jan 23, 2024
Closed
10 tasks
@lasarojc
Copy link
Contributor Author

The other behaviors have not been observed in the latest QA runs, so closing this.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees

@lasarojc lasarojc

Labels
chore documentation Improvements or additions to documentation
Projects
No open projects
CometBFT 2023
Status: Todo
Milestone
No milestone
Development

No branches or pull requests
1 participant
@lasarojc
0