Pin Hyper version to 0.0.26225 - #1308
Conversation
With the new infrastructure introduced in ClickHouse#993 benchmarks will be run via ClickBench-provided infrastructure instead of manually measured. Measurements are kicked off, e.g., by new Pull Requests. This commit pins the Hyper version, such that version bumps in the script will automatically re-trigger the PR-automation remeasuring Hyper's performance numbers. Also see ClickHouse#1266
|
@alexey-milovidov With the automation introduced in #993 , we expected this would run and merge automatically. Can you check why that was not the case? Looks like the GitHub action got stuck.. FYI @rschu1ze |
…ngle (c6a.4xlarge)
|
Results for Logs:
|
|
Results for Logs:
|
…ngle (c6a.2xlarge, c6a.4xlarge, c6a.metal, c7a.metal-48xl)
|
Results for Logs:
|
…quet-single (c6a.2xlarge, c6a.4xlarge, c6a.large, c6a.metal, c6a.xlarge, c7a.metal-48xl)
|
Results for Logs:
|
…ngle (c6a.large, t3a.small)
|
Results for Logs:
|
Umbra was installed and started from umbradb/umbra:latest, so a new image silently changed the reported numbers and nothing re-triggered a measurement. Pinning the tag makes a version bump a change under umbra/, which re-triggers the PR automation, like ClickHouse#1308 did for Hyper (see ClickHouse#1266). 26.08 currently has the same digest as latest, for both amd64 and arm64.
Umbra was installed and started from umbradb/umbra:latest, so a new image silently changed the reported numbers and nothing re-triggered a measurement. Pinning the tag makes a version bump a change under umbra/, which re-triggers the PR automation, like ClickHouse#1308 did for Hyper (see ClickHouse#1266).
With the new infrastructure introduced in #993 benchmarks will be run via ClickBench-provided infrastructure instead of manually measured. Measurements are kicked off, e.g., by new Pull Requests.
This commit pins the Hyper version, such that version bumps in the script will automatically re-trigger the PR-automation remeasuring Hyper's performance numbers.
Also see #1266