chore(workflow): update GHCR cleanup schedule and optimize deletion process

- Changed the cleanup schedule to run hourly for faster package management.
- Adjusted the default dry run mode to false for actual deletions.
- Enhanced logging to provide clearer insights on deletions and remaining versions.
- Implemented limits on deletions per run to avoid rate limits and ensure efficient processing.
This commit is contained in:
zhenlonghe 2025-10-24 14:39:30 +08:00
parent 4c2169610e
commit 75d946e36a
No known key found for this signature in database
GPG Key ID: 9EA159BFE3809CE7
1 changed files with 117 additions and 115 deletions

View File

@ -1,9 +1,9 @@
name: 'Clean up GHCR Packages'
on:
# 每周日 UTC 00:00 执行
# 每小时执行一次(快速清理模式)
schedule:
- cron: '0 0 * * 0'
- cron: '0 * * * *'
# Docker 发布完成后自动触发
workflow_run:
workflows: ['Docker Release']
@ -16,7 +16,7 @@ on:
description: 'Dry run mode (preview only, no actual deletion)'
required: false
type: boolean
default: true
default: false
retention_days:
description: 'Number of days to keep package versions'
required: false
@ -37,7 +37,7 @@ jobs:
- name: Cleanup old GHCR package versions
env:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
DRY_RUN: ${{ github.event.inputs.dry_run || 'true' }}
DRY_RUN: ${{ github.event.inputs.dry_run || 'false' }}
RETENTION_DAYS: ${{ github.event.inputs.retention_days || '30' }}
PACKAGE_NAME: rsshub
OWNER: ${{ github.repository_owner }}
@ -45,24 +45,19 @@ jobs:
set -e
echo "=================================================="
echo "GHCR Package Cleanup (Optimized)"
echo "GHCR Package Cleanup (Fast Mode - Hourly)"
echo "=================================================="
echo "Package: ghcr.io/${OWNER}/${PACKAGE_NAME}"
echo "Retention period: ${RETENTION_DAYS} days"
echo "Dry run mode: ${DRY_RUN}"
echo "Target: ~900 deletions per hour"
echo "Schedule: Every hour (56K versions ~66 hours)"
echo "=================================================="
echo ""
# 计算截止日期UTC时间
if [[ "$OSTYPE" == "darwin"* ]]; then
# macOS
CUTOFF_DATE=$(date -u -v-${RETENTION_DAYS}d +%Y-%m-%dT%H:%M:%SZ)
else
# Linux
CUTOFF_DATE=$(date -u -d "${RETENTION_DAYS} days ago" +%Y-%m-%dT%H:%M:%SZ)
fi
CUTOFF_TIMESTAMP=$(date -d "${CUTOFF_DATE}" +%s 2>/dev/null || date -j -f "%Y-%m-%dT%H:%M:%SZ" "${CUTOFF_DATE}" +%s)
CUTOFF_DATE=$(date -u -d "${RETENTION_DAYS} days ago" +%Y-%m-%dT%H:%M:%SZ)
CUTOFF_TIMESTAMP=$(date -d "${CUTOFF_DATE}" +%s)
echo "Current time (UTC): $(date -u +%Y-%m-%dT%H:%M:%SZ)"
echo "Cutoff date (UTC): ${CUTOFF_DATE}"
@ -70,51 +65,49 @@ jobs:
# 创建临时目录
TEMP_DIR=$(mktemp -d)
TO_DELETE_FILE="${TEMP_DIR}/to_delete.txt"
TO_KEEP_FILE="${TEMP_DIR}/to_keep.txt"
echo "Fetching package versions..."
echo "Fetching package versions (fast mode)..."
# 使用更高效的方式获取版本列表
# 直接使用 gh 命令的 --paginate 选项
# 快速获取前1000个版本最多10页
gh api --paginate \
-H "Accept: application/vnd.github+json" \
-H "X-GitHub-Api-Version: 2022-11-28" \
"/orgs/${OWNER}/packages/container/${PACKAGE_NAME}/versions" > "${TEMP_DIR}/all_versions.json" 2>/dev/null || \
"/orgs/${OWNER}/packages/container/${PACKAGE_NAME}/versions?per_page=100" \
--jq '.[] | select((.created_at | strptime("%Y-%m-%dT%H:%M:%SZ") | mktime) < '"$CUTOFF_TIMESTAMP"') | .id' \
2>/dev/null > "${TEMP_DIR}/to_delete.txt" || \
gh api --paginate \
-H "Accept: application/vnd.github+json" \
-H "X-GitHub-Api-Version: 2022-11-28" \
"/users/${OWNER}/packages/container/${PACKAGE_NAME}/versions" > "${TEMP_DIR}/all_versions.json"
"/users/${OWNER}/packages/container/${PACKAGE_NAME}/versions?per_page=100" \
--jq '.[] | select((.created_at | strptime("%Y-%m-%dT%H:%M:%SZ") | mktime) < '"$CUTOFF_TIMESTAMP"') | .id' \
> "${TEMP_DIR}/to_delete.txt"
TOTAL_VERSIONS=$(jq 'length' "${TEMP_DIR}/all_versions.json")
echo "Total versions found: ${TOTAL_VERSIONS}"
echo ""
TOTAL_TO_DELETE=$(wc -l < "${TEMP_DIR}/to_delete.txt" | tr -d ' ')
echo "Processing versions..."
# 限制每次运行删除数量避免超时保持在1小时内
MAX_PER_RUN=850
# 使用 jq 一次性处理所有版本,生成待删除列表
jq -r --arg cutoff "$CUTOFF_TIMESTAMP" '
.[] |
select((.created_at | strptime("%Y-%m-%dT%H:%M:%SZ") | mktime) < ($cutoff | tonumber)) |
.id
' "${TEMP_DIR}/all_versions.json" > "$TO_DELETE_FILE"
# 生成保留列表
jq -r --arg cutoff "$CUTOFF_TIMESTAMP" '
.[] |
select((.created_at | strptime("%Y-%m-%dT%H:%M:%SZ") | mktime) >= ($cutoff | tonumber)) |
.id
' "${TEMP_DIR}/all_versions.json" > "$TO_KEEP_FILE"
DELETED_COUNT=$(wc -l < "$TO_DELETE_FILE" | tr -d ' ')
KEPT_COUNT=$(wc -l < "$TO_KEEP_FILE" | tr -d ' ')
if [[ $TOTAL_TO_DELETE -gt $MAX_PER_RUN ]]; then
echo "Found ${TOTAL_TO_DELETE} versions to delete"
echo "Limiting to ${MAX_PER_RUN} this run (to stay within rate limits)"
head -n $MAX_PER_RUN "${TEMP_DIR}/to_delete.txt" > "${TEMP_DIR}/to_delete_limited.txt"
mv "${TEMP_DIR}/to_delete_limited.txt" "${TEMP_DIR}/to_delete.txt"
DELETED_COUNT=$MAX_PER_RUN
REMAINING=$((TOTAL_TO_DELETE - MAX_PER_RUN))
else
DELETED_COUNT=$TOTAL_TO_DELETE
REMAINING=0
fi
echo "=================================================="
echo "Analysis Complete"
echo "=================================================="
echo "Total versions: ${TOTAL_VERSIONS}"
echo "Versions to keep: ${KEPT_COUNT}"
echo "Versions to delete: ${DELETED_COUNT}"
echo "Versions to delete this run: ${DELETED_COUNT}"
if [[ $REMAINING -gt 0 ]]; then
echo "Remaining for future runs: ${REMAINING}"
RUNS_NEEDED=$(( (REMAINING + MAX_PER_RUN - 1) / MAX_PER_RUN ))
echo "Estimated additional runs needed: ${RUNS_NEEDED}"
fi
echo ""
if [[ $DELETED_COUNT -eq 0 ]]; then
@ -125,22 +118,12 @@ jobs:
# 显示一些样本
echo "Sample of versions to delete:"
head -n 10 "$TO_DELETE_FILE" | while read -r version_id; do
# 获取版本详情
VERSION_INFO=$(jq -r --arg id "$version_id" '.[] | select(.id == $id)' "${TEMP_DIR}/all_versions.json")
TAGS=$(echo "$VERSION_INFO" | jq -r '.metadata.container.tags[]? // empty' | tr '\n' ',' | sed 's/,$//')
CREATED_AT=$(echo "$VERSION_INFO" | jq -r '.created_at')
AGE_DAYS=$(( ($(date +%s) - $(date -d "$CREATED_AT" +%s 2>/dev/null || date -j -f "%Y-%m-%dT%H:%M:%SZ" "$CREATED_AT" +%s)) / 86400 ))
if [[ -n "$TAGS" ]]; then
echo " Version ID: ${version_id}, Tags: ${TAGS}, Age: ${AGE_DAYS} days"
else
echo " Version ID: ${version_id}, Age: ${AGE_DAYS} days"
fi
head -n 5 "${TEMP_DIR}/to_delete.txt" | while read -r version_id; do
echo " Version ID: ${version_id}"
done
if [[ $DELETED_COUNT -gt 10 ]]; then
echo " ... and $((DELETED_COUNT - 10)) more versions"
if [[ $DELETED_COUNT -gt 5 ]]; then
echo " ... and $((DELETED_COUNT - 5)) more versions"
fi
echo ""
@ -151,81 +134,100 @@ jobs:
exit 0
fi
echo "Starting batch deletion..."
# 批量删除 - 使用并行处理
BATCH_SIZE=50
TOTAL_BATCHES=$(( (DELETED_COUNT + BATCH_SIZE - 1) / BATCH_SIZE ))
echo "Processing ${DELETED_COUNT} versions in ${TOTAL_BATCHES} batches of ${BATCH_SIZE}..."
# 创建批处理文件
split -l $BATCH_SIZE "$TO_DELETE_FILE" "${TEMP_DIR}/batch_"
echo "Starting deletion (optimized for speed)..."
echo "⚠️ Using 4-second intervals (900 requests/hour)"
echo ""
SUCCESS_COUNT=0
ERROR_COUNT=0
PROCESSED=0
START_TIME=$(date +%s)
for batch_file in "${TEMP_DIR}/batch_"*; do
BATCH_NUM=$((SUCCESS_COUNT + ERROR_COUNT + 1))
echo "Processing batch ${BATCH_NUM}/${TOTAL_BATCHES}..."
# 串行删除,带重试机制
while read -r version_id; do
PROCESSED=$((PROCESSED + 1))
RETRY_COUNT=0
MAX_RETRIES=3
SUCCESS=false
# 并行删除当前批次
while read -r version_id; do
(
# 尝试组织 API
while [[ $RETRY_COUNT -lt $MAX_RETRIES ]] && [[ "$SUCCESS" == "false" ]]; do
# 尝试组织 API
DELETE_RESULT=$(gh api \
--method DELETE \
-H "Accept: application/vnd.github+json" \
-H "X-GitHub-Api-Version: 2022-11-28" \
"/orgs/${OWNER}/packages/container/${PACKAGE_NAME}/versions/${version_id}" 2>&1 || echo "ORG_DELETE_FAILED")
# 如果组织 API 失败,尝试用户 API
if [[ "$DELETE_RESULT" == "ORG_DELETE_FAILED" ]] || echo "$DELETE_RESULT" | grep -q "Not Found"; then
DELETE_RESULT=$(gh api \
--method DELETE \
-H "Accept: application/vnd.github+json" \
-H "X-GitHub-Api-Version: 2022-11-28" \
"/orgs/${OWNER}/packages/container/${PACKAGE_NAME}/versions/${version_id}" 2>&1 || echo "ORG_DELETE_FAILED")
# 如果组织 API 失败,尝试用户 API
if [[ "$DELETE_RESULT" == "ORG_DELETE_FAILED" ]] || echo "$DELETE_RESULT" | grep -q "Not Found"; then
DELETE_RESULT=$(gh api \
--method DELETE \
-H "Accept: application/vnd.github+json" \
-H "X-GitHub-Api-Version: 2022-11-28" \
"/users/${OWNER}/packages/container/${PACKAGE_NAME}/versions/${version_id}" 2>&1 || echo "DELETE_FAILED")
fi
if [[ "$DELETE_RESULT" == *"FAILED"* ]] || [[ -n "$DELETE_RESULT" ]]; then
echo "ERROR:${version_id}:${DELETE_RESULT}" >> "${TEMP_DIR}/errors.log"
"/users/${OWNER}/packages/container/${PACKAGE_NAME}/versions/${version_id}" 2>&1 || echo "DELETE_FAILED")
fi
# 检查是否成功
if [[ "$DELETE_RESULT" == *"FAILED"* ]] || [[ -n "$DELETE_RESULT" ]]; then
# 检查是否是速率限制错误
if echo "$DELETE_RESULT" | grep -q "rate limit"; then
RETRY_COUNT=$((RETRY_COUNT + 1))
if [[ $RETRY_COUNT -lt $MAX_RETRIES ]]; then
echo " Rate limit hit, waiting 60s before retry ${RETRY_COUNT}/${MAX_RETRIES}..."
sleep 60
else
echo " Version ${version_id}: Failed after ${MAX_RETRIES} retries (rate limit)"
echo "ERROR:${version_id}:Rate limit exceeded" >> "${TEMP_DIR}/errors.log"
ERROR_COUNT=$((ERROR_COUNT + 1))
fi
else
echo "SUCCESS:${version_id}" >> "${TEMP_DIR}/success.log"
echo " Version ${version_id}: Failed - ${DELETE_RESULT}"
echo "ERROR:${version_id}:${DELETE_RESULT}" >> "${TEMP_DIR}/errors.log"
ERROR_COUNT=$((ERROR_COUNT + 1))
break
fi
) &
done < "$batch_file"
else
SUCCESS=true
SUCCESS_COUNT=$((SUCCESS_COUNT + 1))
fi
done
# 等待当前批次完成
wait
# 显示进度每50个版本
if [[ $((PROCESSED % 50)) -eq 0 ]]; then
CURRENT_TIME=$(date +%s)
ELAPSED=$((CURRENT_TIME - START_TIME))
RATE=$((PROCESSED * 3600 / ELAPSED))
echo "Progress: ${PROCESSED}/${DELETED_COUNT} (Success: ${SUCCESS_COUNT}, Errors: ${ERROR_COUNT}) - Rate: ${RATE}/hour"
fi
# 更新计数
BATCH_SUCCESS=$(wc -l < "${TEMP_DIR}/success.log" 2>/dev/null || echo "0")
BATCH_ERROR=$(wc -l < "${TEMP_DIR}/errors.log" 2>/dev/null || echo "0")
echo " Batch ${BATCH_NUM} completed. Success: ${BATCH_SUCCESS}, Errors: ${BATCH_ERROR}"
done
# 请求间延迟4秒 = 900 requests/hour低于1000限制
sleep 4
done < "${TEMP_DIR}/to_delete.txt"
# 最终统计
FINAL_SUCCESS=$(wc -l < "${TEMP_DIR}/success.log" 2>/dev/null || echo "0")
FINAL_ERROR=$(wc -l < "${TEMP_DIR}/errors.log" 2>/dev/null || echo "0")
END_TIME=$(date +%s)
TOTAL_TIME=$((END_TIME - START_TIME))
ACTUAL_RATE=$((PROCESSED * 3600 / TOTAL_TIME))
echo ""
echo "=================================================="
echo "Cleanup Complete"
echo "=================================================="
echo "Total versions processed: ${TOTAL_VERSIONS}"
echo "Versions kept: ${KEPT_COUNT}"
echo "Versions successfully deleted: ${FINAL_SUCCESS}"
echo "Versions failed to delete: ${FINAL_ERROR}"
echo "Versions successfully deleted: ${SUCCESS_COUNT}"
echo "Versions failed to delete: ${ERROR_COUNT}"
echo "Total time: ${TOTAL_TIME} seconds"
echo "Actual rate: ${ACTUAL_RATE} deletions/hour"
if [[ $FINAL_ERROR -gt 0 ]]; then
if [[ $REMAINING -gt 0 ]]; then
echo ""
echo " Note: This was a partial cleanup run."
echo " Remaining versions will be deleted in the next scheduled run."
echo " Estimated runs needed: $(( (REMAINING + MAX_PER_RUN - 1) / MAX_PER_RUN ))"
fi
if [[ $ERROR_COUNT -gt 0 ]]; then
echo ""
echo "Errors encountered:"
head -n 10 "${TEMP_DIR}/errors.log"
if [[ $FINAL_ERROR -gt 10 ]]; then
echo "... and $((FINAL_ERROR - 10)) more errors"
fi
head -n 5 "${TEMP_DIR}/errors.log" 2>/dev/null || echo "(No error log found)"
fi
echo "=================================================="
@ -233,8 +235,8 @@ jobs:
# 清理临时文件
rm -rf "$TEMP_DIR"
# 如果有错误,退出时返回非零状态码
if [[ $FINAL_ERROR -gt 0 ]]; then
# 如果错误率超过50%,返回失败状态
if [[ $ERROR_COUNT -gt $((SUCCESS_COUNT / 2)) ]]; then
echo "⚠️ High error rate detected"
exit 1
fi
fi