-
Notifications
You must be signed in to change notification settings - Fork 502
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
HDDS-8929. Avoid list allocation for pipeline search #4980
Conversation
Thanks @adoroszlai work on this, this can really improve performance. Although now look at the flame chart, The lock has little impact on latency, and whether the lock will be optimized subsequently, as it can improve throughput. |
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
Thanks @guohao-rosicky, @sodonnel for the review. |
* master: (79 commits) HDDS-8914. Datanode may fail to start due to duplicate VolumeInfoMetrics (apache#4966) HDDS-8921. Add support for EC in Freon SCM block generator (apache#4982) HDDS-8927. Metadata scanner should not scan unhealthy containers. (apache#4976) HDDS-8929. Avoid list allocation for pipeline search (apache#4980) HDDS-8778. Support recursive volume delete using Ozone sh command. (apache#4842) HDDS-8885. Quota repair count enable quota feature for old bucket/volume. (apache#4941) HDDS-8771. Refactor volume level tmp directory for generic usage. (apache#4838) HDDS-8922. Random EC read pipeline ID causes XceiverClient cache churn (apache#4971) HDDS-8586 Recon. - API for Count of deletePending keys and amount of data mapped to such keys. (apache#4923) HDDS-8908. Intermittent failure in TestBlockDeletion#testBlockDeletion (apache#4958) HDDS-8910. Replace LockManager with striped lock in ContainerStateManager (apache#4962) HDDS-8917. Move protobuf conversion out of the lock in PipelineStateManagerImpl (apache#4965) HDDS-8825. Use apache/hadoop 3.3.5 docker image (apache#4963) HDDS-8906. Avoid stream when getting in-service healthy nodes (apache#4960) HDDS-8907. Store volume count when storage report is updated (apache#4957) HDDS-8905. PipelineManager metrics should not be synchronized (apache#4959) HDDS-8553. Improve scanner integration tests. (apache#4936) HDDS-8854. Avoid unnecessary DatanodeDetails creation for NodeStateManager lookup (apache#4925) HDDS-8315. [Snapshot] Added unit tests for SnapshotDiffManager (apache#4716) HDDS-7968. [Snapshot] Improve KeyDeletingService to reclaim eligible key blocks in snapshot's deletedTable (apache#4935) ...
* master: HDDS-8914. Datanode may fail to start due to duplicate VolumeInfoMetrics (apache#4966) HDDS-8921. Add support for EC in Freon SCM block generator (apache#4982) HDDS-8927. Metadata scanner should not scan unhealthy containers. (apache#4976) HDDS-8929. Avoid list allocation for pipeline search (apache#4980) HDDS-8778. Support recursive volume delete using Ozone sh command. (apache#4842) HDDS-8885. Quota repair count enable quota feature for old bucket/volume. (apache#4941) HDDS-8771. Refactor volume level tmp directory for generic usage. (apache#4838) HDDS-8922. Random EC read pipeline ID causes XceiverClient cache churn (apache#4971)
What changes were proposed in this pull request?
Pipeline#equals
includes equality ofgetNodes()
, which wraps the set of nodes in a newArrayList
(defensive "copy"). This is unnecessary, we can compare the underlyingSet
.(see
HealthyPipelineChoosePolicy.choosePipelineIndex
in flamegraph from HDDS-8897)https://issues.apache.org/jira/browse/HDDS-8929
How was this patch tested?
CI:
https://github.com/adoroszlai/hadoop-ozone/actions/runs/5376416993