add prefix heuristics

This commit is contained in:
Bruno Alexandre Rosa 2026-03-14 18:38:08 -03:00
parent 158d9c043e
commit b6bb07f809

View file

@ -1,19 +1,19 @@
#!/bin/bash
# Reports total count and size of GitHub Actions caches whose key starts with a given prefix.
# Used for BuildKit GHA cache (keys like buildkit-blob-1-sha256:...).
# Reports GitHub Actions cache usage: either grouped by key prefix or for a single prefix.
# Prefix = first segment of key (before first "-" or "/"). E.g. buildkit-blob-* -> buildkit, vcpkg/xxx -> vcpkg.
#
# Requires: gh (GitHub CLI), jq
# Run from repo root with gh auth login already done (or GITHUB_TOKEN set in CI).
#
# Usage: .ci/gh_buildkit_cache_usage.sh [key-prefix]
# Default key-prefix: buildkit-blob
# No args: fetch all caches, group by prefix, print table (count + size per prefix) and grand total.
# One arg: report only caches whose key starts with key-prefix (default: buildkit-blob).
#
# Example: .ci/gh_buildkit_cache_usage.sh
# Example: .ci/gh_buildkit_cache_usage.sh buildkit-blob
# Example: .ci/gh_buildkit_cache_usage.sh # all prefixes
# Example: .ci/gh_buildkit_cache_usage.sh buildkit-blob # single prefix
set -eo pipefail
KEY_PREFIX="${1:-buildkit-blob}"
LIMIT=1000
if ! command -v gh &>/dev/null; then
@ -25,6 +25,35 @@ if ! command -v jq &>/dev/null; then
exit 1
fi
if [[ $# -eq 0 ]]; then
# Grouped mode: all caches, group by first segment (before - or /)
json=$(gh cache list --limit "$LIMIT" --json key,sizeInBytes 2>/dev/null) || {
echo "Error: gh cache list failed (check gh auth or repo access)" >&2
exit 1
}
# Add prefix field, group by prefix, sum count and size, sort by size desc; output GiB (2 decimals)
total_entries=$(echo "$json" | jq 'length')
total_bytes=$(echo "$json" | jq 'map(.sizeInBytes) | add // 0')
total_gib=$(echo "$total_bytes" | jq -r '. / 1024 / 1024 / 1024 | . * 100 | floor / 100')
printf "%-12s %6s %10s\n" "PREFIX" "COUNT" "SIZE (GiB)"
printf "%-12s %6s %10s\n" "------------" "------" "----------"
while IFS=$'\t' read -r prefix count gib; do
printf "%-12s %6s %10.2f\n" "$prefix" "$count" "$gib"
done < <(echo "$json" | jq -r '
def prefix: (if (.key | test("/")) then (.key | split("/")[0]) else (.key | split("-")[0]) end);
def to_gib: . / 1024 / 1024 / 1024 | . * 100 | floor / 100;
map(. + {prefix: prefix}) | group_by(.prefix) |
map({prefix: .[0].prefix, count: length, totalBytes: (map(.sizeInBytes) | add // 0)}) |
sort_by(-.totalBytes) |
.[] | "\(.prefix)\t\(.count)\t\(.totalBytes | to_gib)"
')
echo ""
printf "Total: %s entries, %s GiB\n" "$total_entries" "$total_gib"
exit 0
fi
# Single-prefix mode
KEY_PREFIX="${1:-buildkit-blob}"
json=$(gh cache list --key "$KEY_PREFIX" --limit "$LIMIT" --json key,sizeInBytes 2>/dev/null) || {
echo "Error: gh cache list failed (check gh auth or repo access)" >&2
exit 1
@ -32,8 +61,6 @@ json=$(gh cache list --key "$KEY_PREFIX" --limit "$LIMIT" --json key,sizeInBytes
count=$(echo "$json" | jq 'length')
total_bytes=$(echo "$json" | jq 'map(.sizeInBytes) | add // 0')
# Human-readable size (GiB, 2 decimals)
total_gib=$(echo "$total_bytes" | jq -r '. / 1024 / 1024 / 1024 | . * 100 | floor / 100')
echo "Cache key prefix: $KEY_PREFIX"