-
Notifications
You must be signed in to change notification settings - Fork 1.7k
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
chore(deps): update #6217
chore(deps): update #6217
Conversation
Fix for |
Looks like |
I think Jesse disable benchmark fails, but maybe I'm wrong. Regressed benchmarks:
|
cc @MOZGIII on the regression since you were working on performance in that crate. |
The results suggest that the system performance is unstable. The benches are written in such a way that they're supposed to highlight overhead from certain operations, and in this case the regression can be observed in the benches that are 100% unaffected by the metrics crate - I'd prefer to get a second opinion on this, but I think it's the system instability that caused this numbers. I'll double-check the source to reassure. |
Week update.
|
I'm not sure if it is related to the
I'm running this on of the benchmark runners to reduce noise. Simply running the master benchmarks twice, for example, does not show regressions as might be expected if they were noisy. I'll try to run the full suite. |
To answer the question about the CI run, we haven't yet enabled failing that workflow when regressions are detected. I wanted to collect some data about how often we see false positives with it now that it is mostly stable. |
Or rather, we have the results from the last CI run. The changed benchmarks:
It seems like the bulk are in the metrics subsystem micro-benches. I think it is worth looking at them more closely; maybe starting with the changelog for the metrics crate. |
Can we remove the metrics upgrade and proceed with upgrading the rest of the dependencies? |
Cargo.toml
Outdated
dashmap = "3" | ||
dashmap = "4" |
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.
ISTR there was a regression issue with dashmap 4, though I don't recall the details. Perhaps relevant to the metrics performance regression?
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.
It is exactly my thinking as well - dashmap
4 has a different and slower implementation, and I suspect this to be the cause of the regression.
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.
Downgrade dashmap
back to 3.
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.
@fanatid do you think this was the cause of the regression you noted here? If so, we might be able to upgrade the metrics
dependency?
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.
No, it does not help. I downgraded metrics crates in addition and waiting benchmark for confirmation that everything is ok.
801d39c
to
950630f
Compare
Still regression for some metrics benchmarks:
|
To clarify, removing the |
Yes, that's correct, I removed |
Signed-off-by: Kirill Fomichev <fanatid@ya.ru>
Signed-off-by: Kirill Fomichev <fanatid@ya.ru>
00caf59
to
06c8c8b
Compare
Signed-off-by: Kirill Fomichev <fanatid@ya.ru>
Signed-off-by: Kirill Fomichev <fanatid@ya.ru>
Signed-off-by: Kirill Fomichev <fanatid@ya.ru>
I pinned metrics crates, no regression now. CI job: https://github.com/timberio/vector/runs/1869218472?check_suite_focus=true Updated crates (in
|
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.
Cool, looks good!
Signed-off-by: Kirill Fomichev <fanatid@ya.ru>
It was pinned in #6217 because it was believed it required Tokio 1.0 but I'm not actually seeing this. Signed-off-by: Jesse Szwedko <jesse@szwedko.me>
metrics
crates chore(deps): bump metrics-util from 0.4.0-alpha.10 to 0.4.0-alpha.11 #6206rust_decimal
chore(deps): bump rust_decimal from 1.9.0 to 1.10.1 #6213cargo update --aggressive
I also pinned
async-graphql
/async-graphql-warp
because they addtokio:1.0
(#5175).