-
Notifications
You must be signed in to change notification settings - Fork 659
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
Atlas fixes #2453
Atlas fixes #2453
Conversation
Pretty happy with the result I'm seeing on my local setup, simulating 2000 users pre-ordering + registering names, so a total of 6000 transactions being dispatched on 10 follower nodes and 1 miner, involving attachments replication. |
Notes for performance tests:
|
47c1536
to
6d66390
Compare
I've spent some time benchmarking performances on my local machine, and with:
|
Same setup and same miner, after killing and restarting the 9 followers a few times:
So zooming on the Atlas histograms:
and
We're meeting the acceptance criteria (no response time > 100ms):
|
Thanks for gathering the data! How big were these inventories in these tests? Were they maximally sized -- i.e. is this data the worst-case possible performance? |
Got a panic while testing this. I stopped my testnet node and restarted it, and it immediately crashed with this trace:
I did get this warning from Atlas just before the crash: |
@jcnelson thanks for reporting, I think I see what's going on, this is with a |
Yes, there's a |
b14bef7
to
cdebac5
Compare
@jcnelson bug was fixed, the latest branch is doing a better job at dealing with graceful working dir recovery. |
Found another potential problem -- Atlas attachment instances are inserted multiple times across node re-starts:
Is this supposed to happen? Also, I'm running a testnet follower on Xenon, and it's not getting any attachments at all. Does the Xenon node have attachments to serve, or should I go and make some? |
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! Once this deploys to Xenon, let's register a bunch of names with attachments (e.g. on the order of 1,000) and verify that the network remains stable.
# Conflicts: # src/monitoring/mod.rs # src/monitoring/prometheus.rs
@jcnelson Sounds good, will do. |
Addressing #2428, #2427 and #2425.
I'm still working on some tooling for #2142, which I believe could help revealing unexpected situations.