Open
Implement blobstore-level metrics collection and reporting #9
wuxingyiAdd on  Dec 9, 2024
wuxingyiCollaborators
Dec 9, 2024 Create

1.Description:
Implement blobstore-level metrics collection and reporting to monitor service, following the existing monitoring architecture.

2.Technical Requirements:
Collect detailed blobstore metrics:

struct BlobstoreMetrics {
    // IO Statistics
    struct IOStats {
        uint64_t read_ops;
        uint64_t write_ops;
        uint64_t read_bytes;
        uint64_t write_bytes;
        uint64_t read_latency_us;  // Average
        uint64_t write_latency_us; // Average
    } io_stats;

    // Blob Statistics
    struct BlobStats {
        uint64_t total_blobs;
        uint64_t active_blobs;
        uint64_t deleted_blobs;
        uint64_t total_clusters;
        uint64_t free_clusters;
    } blob_stats;

    // Error Statistics
    struct ErrorStats {
        uint64_t read_errors;
        uint64_t write_errors;
        uint64_t checksum_errors;
        uint64_t metadata_errors;
    } error_stats;
};
  1. Expected Results:
    a. Metrics Collected:

    • [IO operations (read/write counts, bytes, latency)
    • Blob statistics (counts, space usage)
    • Error counts and types

    b. Reporting:

    • Regular reporting to monitor with other OSD stats
    • Per-blobstore granularity
    • Aggregated OSD-level view
  2. Acceptance Criteria:

  • Metrics collection adds < 1% overhead to operations
  • All critical metrics are captured
  • Proper integration with existing OSD reporting
  • Metrics successfully received by monitor
  • Documentation for all available metrics
likedislike
WwuxingyiCollaborators
Dec 9, 2024 创建了任务
openeuler-ci-botopeneuler-ci-botCollaborators
Dec 9, 2024 添加了 &nbsp; sig/sig-SDS 标签
WwuxingyiCollaborators
Dec 9, 2024 修改了描述