List Regressions

List Regressions

This skill provides functionality to fetch regression data for OpenShift components across different releases. It uses a Python script to query a component health API and retrieve regression information.

When to Use This Skill

Use this skill when you need to:

• Analyze component health for a specific OpenShift release
• Track regressions across releases
• Filter regressions by their open/closed status
• Generate reports on component stability

Prerequisites

1. Python 3 Installation

   • Check if installed: which python3
   • Python 3.6 or later is required
   • Comes pre-installed on most systems

2. Network Access

   • The script requires network access to reach the component health API
   • Ensure you can make HTTPS requests

3. API Endpoint Configuration

   • The script includes a placeholder API endpoint that needs to be updated
   • Update the base_url in list_regressions.py with the actual component health API endpoint

Implementation Steps

Step 1: Verify Prerequisites

First, ensure Python 3 is available:

python3 --version

If Python 3 is not installed, guide the user through installation for their platform.

Step 2: Locate the Script

The script is located at:

plugins/teams/skills/list-regressions/list_regressions.py

Step 3: Run the Script

Execute the script with appropriate arguments:

# Basic usage - all regressions for a release
python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.17

# Filter by specific components
python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.21 \
  --components Monitoring "kube-apiserver"

# Filter by multiple components
python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.21 \
  --components Monitoring etcd "kube-apiserver"

# Filter by development window (GA'd release - both start and end)
python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.17 \
  --start 2024-05-17 \
  --end 2024-10-01

# Filter by development window (in-development release - start only)
python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.21 \
  --start 2025-09-02

Step 4: Process the Output

The script outputs JSON data with the following structure:

{
  "summary": {
    "total": <number>,
    "triaged": <number>,
    "triage_percentage": <number>,
    "time_to_triage_hrs_avg": <number or null>,
    "time_to_triage_hrs_max": <number or null>,
    "time_to_resolve_hrs_avg": <number or null>,
    "time_to_resolve_hrs_max": <number or null>,
    "open": {
      "total": <number>,
      "triaged": <number>,
      "triage_percentage": <number>,
      "time_to_triage_hrs_avg": <number or null>,
      "time_to_triage_hrs_max": <number or null>,
      "open_hrs_avg": <number or null>,
      "open_hrs_max": <number or null>
    },
    "closed": {
      "total": <number>,
      "triaged": <number>,
      "triage_percentage": <number>,
      "time_to_triage_hrs_avg": <number or null>,
      "time_to_triage_hrs_max": <number or null>,
      "time_to_resolve_hrs_avg": <number or null>,
      "time_to_resolve_hrs_max": <number or null>
    }
  },
  "components": {
    "ComponentName": {
      "summary": {
        "total": <number>,
        "triaged": <number>,
        "triage_percentage": <number>,
        "time_to_triage_hrs_avg": <number or null>,
        "time_to_triage_hrs_max": <number or null>,
        "time_to_resolve_hrs_avg": <number or null>,
        "time_to_resolve_hrs_max": <number or null>,
        "open": {
          "total": <number>,
          "triaged": <number>,
          "triage_percentage": <number>,
          "time_to_triage_hrs_avg": <number or null>,
          "time_to_triage_hrs_max": <number or null>,
          "open_hrs_avg": <number or null>,
          "open_hrs_max": <number or null>
        },
        "closed": {
          "total": <number>,
          "triaged": <number>,
          "triage_percentage": <number>,
          "time_to_triage_hrs_avg": <number or null>,
          "time_to_triage_hrs_max": <number or null>,
          "time_to_resolve_hrs_avg": <number or null>,
          "time_to_resolve_hrs_max": <number or null>
        }
      },
      "open": [...],
      "closed": [...]
    }
  }
}

CRITICAL: The output includes pre-calculated counts and health metrics:

• summary: Overall statistics across all components
  • summary.total: Total number of regressions
  • summary.triaged: Total number of regressions triaged (open + closed)
  • summary.triage_percentage: Percentage of all regressions that have been triaged (KEY HEALTH METRIC)
  • summary.time_to_triage_hrs_avg: Overall average hours to triage (combining open and closed, KEY HEALTH METRIC)
  • summary.time_to_triage_hrs_max: Overall maximum hours to triage
  • summary.time_to_resolve_hrs_avg: Overall average hours to resolve regressions, measured from regression opened to triage resolved timestamp (KEY HEALTH METRIC)
  • summary.time_to_resolve_hrs_max: Overall maximum hours to resolve regressions
  • summary.open.total: Number of open regressions (where closed is null)
  • summary.open.triaged: Number of open regressions that have been triaged to a JIRA bug
  • summary.open.triage_percentage: Percentage of open regressions triaged
  • summary.open.time_to_triage_hrs_avg: Average hours from opened to first triage (open only)
  • summary.open.time_to_triage_hrs_max: Maximum hours from opened to first triage (open only)
  • summary.open.open_hrs_avg: Average hours that open regressions have been open (from opened to current time)
  • summary.open.open_hrs_max: Maximum hours that open regressions have been open (from opened to current time)
  • summary.closed.total: Number of closed regressions (where closed is not null)
  • summary.closed.triaged: Number of closed regressions that have been triaged to a JIRA bug
  • summary.closed.triage_percentage: Percentage of closed regressions triaged
  • summary.closed.time_to_triage_hrs_avg: Average hours from opened to first triage (closed only)
  • summary.closed.time_to_triage_hrs_max: Maximum hours from opened to first triage (closed only)
  • summary.closed.time_to_resolve_hrs_avg: Average hours from regression opened to triage resolved (JIRA bug completed), null if no valid data
  • summary.closed.time_to_resolve_hrs_max: Maximum hours from regression opened to triage resolved, null if no valid data
• components: Dictionary mapping component names to objects containing:
  • summary: Per-component statistics (includes same fields as overall summary)
  • open: Array of open regression objects for that component
  • closed: Array of closed regression objects for that component

Time to Triage Calculation:

The time_to_triage_hrs_avg field is calculated as:

1. For each triaged regression, find the earliest created_at timestamp in the triages array
2. Calculate the time difference between the regression's opened timestamp and the earliest triage timestamp
3. Convert the difference to hours and round to the nearest hour
4. Only include positive time differences (zero or negative values are skipped - these occur when triages are reused across regression instances)
5. Average all valid time-to-triage values for open regressions separately from closed regressions
6. Return null if no regressions have valid time-to-triage data in that category

Time to Resolve Calculation:

The time_to_resolve_hrs_avg and time_to_resolve_hrs_max fields (only for triaged closed regressions) are calculated as:

1. For each closed regression that has been triaged, find the earliest resolved timestamp among the triage objects
2. The triage resolved timestamp indicates when the linked JIRA bug moved to a completed state
3. Calculate the time difference between the regression's opened timestamp and the earliest triage resolved timestamp
4. Convert the difference to hours and round to the nearest hour
5. Only include positive time differences (skip data inconsistencies)
6. Calculate average and maximum of all valid time-to-resolve values
7. Return null if no triaged closed regressions have valid resolved timestamps

Note: This metric uses the triage resolved timestamp rather than the regression closed timestamp because the regression closing (bad data rolling off component readiness) can lag by a week or more after the underlying issue is fixed. The triage resolved timestamp provides a more accurate measure of how quickly a team addressed the issue.

Open Duration Calculation:

The open_hrs_avg and open_hrs_max fields (only for open regressions) are calculated as:

1. For each open regression, calculate the time difference between opened timestamp and current time
2. Convert the difference to hours and round to the nearest hour
3. Only include positive time differences
4. Calculate average and maximum of all open duration values
5. Return null if no open regressions have valid time data

ALWAYS use these summary counts rather than attempting to count the regression arrays yourself. This ensures accuracy even when the output is truncated due to size.

The script automatically simplifies and optimizes the response:

Time field simplification (closed and last_failure):

• Original API format: {"Time": "2025-09-27T12:04:24.966914Z", "Valid": true}
• Simplified format: "closed": "2025-09-27T12:04:24.966914Z" (if Valid is true)
• Or: "closed": null (if Valid is false)
• Same applies to last_failure field

Field removal for response size optimization:

• links: Removed from each regression (reduces response size significantly)
• test_id: Removed from each regression (large field, can be reconstructed from test_name if needed)

Date filtering (optional):

• Use --start and --end parameters to filter regressions to a specific time window
• --start YYYY-MM-DD: Excludes regressions that were closed before this date
• --end YYYY-MM-DD: Excludes regressions that were opened after this date
• Typical use case: Filter to the development window
  • --start: development_start date from get-release-dates skill (always applied)
  • --end: GA date from get-release-dates skill (only for GA'd releases)
• For GA'd releases: Both start and end filtering applied
• For in-development releases (null GA date): Only start filtering applied (no end date)
• Benefits: Focuses analysis on regressions during active development, excluding:
  • Regressions closed before the release development started (not relevant)
  • Regressions opened after GA (post-release, often not monitored/triaged - GA'd releases only)

Parse this JSON output to extract relevant information for analysis.

Step 5: Generate Analysis (Optional)

Based on the regression data:

1. Use the summary counts from the summary and components.*.summary objects (do NOT count the arrays)
2. Identify most affected components using components.*.summary.open.total
3. Compare with previous releases
4. Analyze trends in open vs closed regressions per component
5. Create visualizations if needed

Error Handling

Common Errors

1. Network Errors

   • Symptom: URLError or connection timeout
   • Solution: Check network connectivity and firewall rules
   • Retry: The script has a 30-second timeout, consider retrying

2. HTTP Errors

   • Symptom: HTTP 404, 500, etc.
   • Solution: Verify the API endpoint URL is correct
   • Check: Ensure the release parameter is valid

3. Invalid Release

   • Symptom: Empty results or error response
   • Solution: Verify the release format (e.g., "4.17", not "v4.17")

4. Invalid Boolean Value

   • Symptom: ValueError: Invalid boolean value
   • Solution: Use only "true" or "false" for the --opened flag

Debugging

Enable verbose output by examining stderr:

python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.17 2>&1 | tee debug.log

Script Arguments

Required Arguments

• --release: Release version to query
  • Format: "X.Y" (e.g., "4.17", "4.16")
  • Must be a valid OpenShift release number

Optional Arguments

• --components: Filter by component names

  • Values: Space-separated list of component names
  • Default: None (returns all components)
  • Case-insensitive matching
  • Examples: --components Monitoring etcd "kube-apiserver"
  • Filtering is performed after fetching data from the API

• --test-name: Filter by exact test name

  • Value: Exact test name string (case-sensitive)
  • Returns only regressions for this specific test across all variants and components
  • Useful for finding all variant-specific regressions for a single test
  • Example: --test-name "[Monitor:kubelet-container-restarts][sig-architecture] platform pods in ns/openshift-machine-config-operator should not exit an excessive amount of times"

• --test-name-contains: Filter by test name substring

  • Value: Substring to search for within test names (case-insensitive)
  • Returns regressions whose test name contains the given string
  • Useful for finding related tests in the same namespace or sig
  • Example: --test-name-contains "openshift-machine-config-operator"

Output Format

The script outputs JSON with summaries and regressions grouped by component:

{
  "summary": {
    "total": 62,
    "triaged": 59,
    "triage_percentage": 95.2,
    "time_to_triage_hrs_avg": 68,
    "time_to_triage_hrs_max": 240,
    "time_to_resolve_hrs_avg": 168,
    "time_to_resolve_hrs_max": 480,
    "open": {
      "total": 2,
      "triaged": 1,
      "triage_percentage": 50.0,
      "time_to_triage_hrs_avg": 48,
      "time_to_triage_hrs_max": 48,
      "open_hrs_avg": 120,
      "open_hrs_max": 200
    },
    "closed": {
      "total": 60,
      "triaged": 58,
      "triage_percentage": 96.7,
      "time_to_triage_hrs_avg": 72,
      "time_to_triage_hrs_max": 240,
      "time_to_resolve_hrs_avg": 168,
      "time_to_resolve_hrs_max": 480
    }
  },
  "components": {
    "Monitoring": {
      "summary": {
        "total": 15,
        "triaged": 13,
        "triage_percentage": 86.7,
        "time_to_triage_hrs_avg": 68,
        "time_to_triage_hrs_max": 180,
        "time_to_resolve_hrs_avg": 156,
        "time_to_resolve_hrs_max": 360,
        "open": {
          "total": 1,
          "triaged": 0,
          "triage_percentage": 0.0,
          "time_to_triage_hrs_avg": null,
          "time_to_triage_hrs_max": null,
          "open_hrs_avg": 72,
          "open_hrs_max": 72
        },
        "closed": {
          "total": 14,
          "triaged": 13,
          "triage_percentage": 92.9,
          "time_to_triage_hrs_avg": 68,
          "time_to_triage_hrs_max": 180,
          "time_to_resolve_hrs_avg": 156,
          "time_to_resolve_hrs_max": 360
        }
      },
      "open": [
        {
          "id": 12894,
          "component": "Monitoring",
          "closed": null,
          ...
        }
      ],
      "closed": [
        {
          "id": 12893,
          "view": "4.21-main",
          "release": "4.21",
          "base_release": "4.18",
          "component": "Monitoring",
          "capability": "operator-conditions",
          "test_name": "...",
          "variants": [...],
          "opened": "2025-09-26T00:02:51.385944Z",
          "closed": "2025-09-27T12:04:24.966914Z",
          "triages": [],
          "last_failure": "2025-09-25T14:41:17Z",
          "max_failures": 9
        }
      ]
    },
    "etcd": {
      "summary": {
        "total": 20,
        "triaged": 19,
        "triage_percentage": 95.0,
        "time_to_triage_hrs_avg": 84,
        "time_to_triage_hrs_max": 220,
        "time_to_resolve_hrs_avg": 192,
        "time_to_resolve_hrs_max": 500,
        "open": {
          "total": 0,
          "triaged": 0,
          "triage_percentage": 0.0,
          "time_to_triage_hrs_avg": null,
          "time_to_triage_hrs_max": null,
          "open_hrs_avg": null,
          "open_hrs_max": null
        },
        "closed": {
          "total": 20,
          "triaged": 19,
          "triage_percentage": 95.0,
          "time_to_triage_hrs_avg": 84,
          "time_to_triage_hrs_max": 220,
          "time_to_resolve_hrs_avg": 192,
          "time_to_resolve_hrs_max": 500
        }
      },
      "open": [],
      "closed": [...]
    },
    "kube-apiserver": {
      "summary": {
        "total": 27,
        "triaged": 27,
        "triage_percentage": 100.0,
        "time_to_triage_hrs_avg": 58,
        "time_to_triage_hrs_max": 168,
        "time_to_resolve_hrs_avg": 144,
        "time_to_resolve_hrs_max": 400,
        "open": {
          "total": 1,
          "triaged": 1,
          "triage_percentage": 100.0,
          "time_to_triage_hrs_avg": 36,
          "time_to_triage_hrs_max": 36,
          "open_hrs_avg": 96,
          "open_hrs_max": 96
        },
        "closed": {
          "total": 26,
          "triaged": 26,
          "triage_percentage": 100.0,
          "time_to_triage_hrs_avg": 60,
          "time_to_triage_hrs_max": 168,
          "time_to_resolve_hrs_avg": 144,
          "time_to_resolve_hrs_max": 400
        }
      },
      "open": [...],
      "closed": [...]
    }
  }
}

Important - Summary Objects:

• The summary object contains overall pre-calculated counts for accuracy
• Each component in the components object has its own summary with per-component counts
• The components object maps component names (sorted alphabetically) to objects containing:
  • summary: Statistics for this component (total, open, closed)
  • open: Array of open regression objects (where closed is null)
  • closed: Array of closed regression objects (where closed has a timestamp)
• ALWAYS use the summary and components.*.summary fields for counts (including total, open.total, open.triaged, closed.total, closed.triaged)
• Do NOT attempt to count the components.*.open or components.*.closed arrays yourself

Note: Time fields are simplified from the API response:

• closed: If the regression is closed: "closed": "2025-09-27T12:04:24.966914Z" (timestamp string), otherwise null
• last_failure: If valid: "last_failure": "2025-09-25T14:41:17Z" (timestamp string), otherwise null

Examples

Example 1: List All Regressions

python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.17

Expected Output: JSON containing all regressions for release 4.17

Example 2: Filter by Component

python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.21 \
  --components Monitoring etcd

Expected Output: JSON containing regressions for only Monitoring and etcd components in release 4.21

Example 3: Filter by Single Component

python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.21 \
  --components "kube-apiserver"

Expected Output: JSON containing regressions for the kube-apiserver component in release 4.21

Example 4: Filter by Exact Test Name

python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.22 \
  --test-name "[Monitor:kubelet-container-restarts][sig-architecture] platform pods in ns/openshift-machine-config-operator should not exit an excessive amount of times"

Expected Output: JSON containing all regressions (across all components and variants) for this exact test in release 4.22. Useful for finding the same test failing in different variant combinations.

Example 5: Filter by Test Name Substring

python3 plugins/teams/skills/list-regressions/list_regressions.py \
  --release 4.22 \
  --test-name-contains "openshift-machine-config-operator"

Expected Output: JSON containing all regressions whose test name contains the substring (case-insensitive). Useful for finding related tests in the same namespace or sig.

Note: --test-name and --test-name-contains are mutually exclusive with each other and with --components/--team. They search across all components automatically.

Customization

Updating the API Endpoint

The script includes a placeholder API endpoint. Update it in list_regressions.py:

# Current placeholder
base_url = f"https://teams-api.example.com/api/v1/regressions"

# Update to actual endpoint
base_url = f"https://actual-api.example.com/api/v1/regressions"

Adding Custom Filters

To add additional query parameters, modify the fetch_regressions function:

def fetch_regressions(release: str, opened: Optional[bool] = None,
                     component: Optional[str] = None) -> dict:
    params = [f"release={release}"]
    if opened is not None:
        params.append(f"opened={'true' if opened else 'false'}")
    if component is not None:
        params.append(f"component={component}")
    # ... rest of function

Integration with Commands

This skill is designed to be used by the /teams:analyze-regressions command, but can also be invoked directly by other commands or scripts that need regression data.

Related Skills

• Component health analysis
• Release comparison
• Regression tracking
• Quality metrics reporting

Notes

• The script uses Python's built-in urllib module (no external dependencies)
• Output is always JSON format for easy parsing
• Diagnostic messages are written to stderr, data to stdout
• The script has a 30-second timeout for HTTP requests