Improve SKU catalog hygiene and BOM accuracy
All checks were successful
Deploy Skill to OCI / deploy (push) Successful in 21s

- refresh_sku_catalog.py: add --discover to report API SKUs missing from
  catalog, filtered to already-curated serviceCategory values. Auto-runs
  at end of --validate and --refresh so gaps surface on every maintenance
  pass. First run found 162 missing SKUs (Blackwell GPUs, X12 Ax compute,
  VMware reserved tiers, WebLogic-on-OKE, Analytics/OIC BYOL).
- oci_bom_gen.py: fix Cost % column being blank. Back-fill was gated on
  formula detection but data rows write raw numbers, so the predicate
  never matched. Now tracks data_item_rows explicitly.
- .gitea/workflows/sku-catalog-refresh.yaml: monthly automation (1st at
  09:00 UTC). Auto-refresh prices → push branch + open PR; detect new
  SKUs → open issue with labels. Gated on secrets.GITEA_TOKEN.
- kb/pricing/oci-sku-catalog.yaml: add B95714 / B95715 (Autonomous ATP
  Dedicated ECPU, LI + BYOL) — canonical SKUs for ADB-D pricing.
- kb/services/compute.yaml: add X12 family (VM.Standard4.Ax.Flex,
  BM.Standard4.Ax.120), E6.Ax and A4.Ax variants. Verified against
  docs.oracle.com computeshapes.htm. Bump last_verified + changelog.
- templates/bom-spec.yaml: document 730 hrs/month convention as default
  (= real annual billing / 12; 744 overstates by 1.92% on 12-month TCO).
- Makefile: new 'make sku-discover' target.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
root
2026-04-24 11:40:53 -03:00
parent a22711c065
commit 874d40d38a
7 changed files with 327 additions and 13 deletions

View File

@@ -0,0 +1,143 @@
# Monthly SKU catalog refresh.
#
# Runs on the 1st of each month at 09:00 UTC and on manual dispatch.
#
# Does two things:
# 1. `--refresh`: pulls latest PAYG prices from Oracle's public API,
# updates kb/pricing/oci-sku-catalog.yaml, and opens a PR if anything
# changed.
# 2. `--discover`: reports SKUs present in the API but absent from the
# catalog (filtered to serviceCategory values already curated).
# If new SKUs are found, opens a Gitea issue listing them by category
# for manual review.
#
# Tokens:
# secrets.GITEA_TOKEN — must have repo:write + issues:write.
# If not set, the run logs the diff/discover output as an artifact and
# leaves the rest (branch push, PR, issue) as a no-op.
name: SKU Catalog Refresh
on:
schedule:
- cron: '0 9 1 * *'
workflow_dispatch:
jobs:
refresh-and-discover:
runs-on: ubuntu-latest
timeout-minutes: 15
steps:
- name: Checkout
uses: actions/checkout@v4
with:
fetch-depth: 0
- name: Setup Python
run: |
sudo apt-get update -qq
sudo apt-get install -y python3 python3-pip jq
python3 -m pip install --user requests pyyaml
- name: Refresh prices from Oracle API
id: refresh
run: |
python3 tools/refresh_sku_catalog.py --refresh --diff 2>&1 \
| tee /tmp/refresh.log
if git diff --quiet kb/pricing/oci-sku-catalog.yaml; then
echo "changed=false" >> $GITHUB_OUTPUT
else
echo "changed=true" >> $GITHUB_OUTPUT
fi
- name: Discover missing SKUs
id: discover
run: |
python3 tools/refresh_sku_catalog.py --discover -v 2>&1 \
| tee /tmp/discover.log
NEW_COUNT=$(grep -oE '^[0-9]+ SKUs present in API' /tmp/discover.log \
| head -1 | awk '{print $1}' || echo "0")
echo "new_count=${NEW_COUNT:-0}" >> $GITHUB_OUTPUT
- name: Upload logs as artifact
if: always()
uses: actions/upload-artifact@v4
with:
name: sku-refresh-logs
path: |
/tmp/refresh.log
/tmp/discover.log
retention-days: 90
- name: Open PR with price updates
if: steps.refresh.outputs.changed == 'true' && env.GITEA_TOKEN != ''
env:
GITEA_TOKEN: ${{ secrets.GITEA_TOKEN }}
GITEA_SERVER: ${{ github.server_url }}
GITEA_REPO: ${{ github.repository }}
run: |
DATE=$(date -u +%Y-%m-%d)
BRANCH="automation/sku-refresh-${DATE}"
git config user.name "sku-refresh-bot"
git config user.email "sku-refresh@automation.local"
git checkout -b "$BRANCH"
git add kb/pricing/oci-sku-catalog.yaml
git commit -m "chore(pricing): monthly SKU catalog refresh (${DATE})
Automated refresh from Oracle public pricing API.
See workflow artifact 'sku-refresh-logs' for the diff."
# Push using token auth
REMOTE=$(echo "$GITEA_SERVER/$GITEA_REPO.git" \
| sed "s#https://#https://sku-refresh-bot:${GITEA_TOKEN}@#")
git push "$REMOTE" "$BRANCH"
# Open PR via Gitea API
OWNER=$(echo "$GITEA_REPO" | cut -d/ -f1)
REPO=$(echo "$GITEA_REPO" | cut -d/ -f2)
curl -sS -X POST \
-H "Authorization: token ${GITEA_TOKEN}" \
-H "Content-Type: application/json" \
"${GITEA_SERVER}/api/v1/repos/${OWNER}/${REPO}/pulls" \
-d "$(jq -n \
--arg title "chore(pricing): monthly SKU refresh ${DATE}" \
--arg head "$BRANCH" \
--arg base "main" \
--arg body "$(printf 'Automated refresh — [refresh log artifact](../../actions).\n\n```\n%s\n```' \
"$(tail -40 /tmp/refresh.log)")" \
'{title:$title, head:$head, base:$base, body:$body}')"
- name: Open issue for new SKUs
if: steps.discover.outputs.new_count != '0' && env.GITEA_TOKEN != ''
env:
GITEA_TOKEN: ${{ secrets.GITEA_TOKEN }}
GITEA_SERVER: ${{ github.server_url }}
GITEA_REPO: ${{ github.repository }}
NEW_COUNT: ${{ steps.discover.outputs.new_count }}
run: |
DATE=$(date -u +%Y-%m-%d)
OWNER=$(echo "$GITEA_REPO" | cut -d/ -f1)
REPO=$(echo "$GITEA_REPO" | cut -d/ -f2)
BODY=$(printf '## %s new SKUs detected in Oracle API\n\n**Detected:** %s\n\nThese SKUs are present in the Oracle public pricing API but not yet in `kb/pricing/oci-sku-catalog.yaml`. Filter: serviceCategory values already represented in our catalog (so only relevant families show up).\n\n### Full listing\n\n```\n%s\n```\n\n### Next step\n\nReview and add relevant entries to `kb/pricing/oci-sku-catalog.yaml` under the appropriate category block. After merging, prices will be picked up automatically on the next monthly refresh.\n' \
"$NEW_COUNT" "$DATE" "$(cat /tmp/discover.log)")
curl -sS -X POST \
-H "Authorization: token ${GITEA_TOKEN}" \
-H "Content-Type: application/json" \
"${GITEA_SERVER}/api/v1/repos/${OWNER}/${REPO}/issues" \
-d "$(jq -n \
--arg title "SKU catalog: ${NEW_COUNT} new SKUs to review (${DATE})" \
--arg body "$BODY" \
'{title:$title, body:$body, labels:["automation","sku-catalog"]}')"
- name: Summary
if: always()
run: |
echo "## SKU Refresh Summary" >> $GITHUB_STEP_SUMMARY
echo "- Catalog changed: ${{ steps.refresh.outputs.changed }}" >> $GITHUB_STEP_SUMMARY
echo "- New SKUs discovered: ${{ steps.discover.outputs.new_count }}" >> $GITHUB_STEP_SUMMARY
echo "- Token configured: ${{ env.GITEA_TOKEN != '' && 'yes' || 'no (PR/issue skipped)' }}" >> $GITHUB_STEP_SUMMARY
env:
GITEA_TOKEN: ${{ secrets.GITEA_TOKEN }}

View File

@@ -1,6 +1,6 @@
# OCI Deal Accelerator — Build Automation # OCI Deal Accelerator — Build Automation
.PHONY: help install test validate example diagram deck full clean lint codex-package update-icons freshness freshness-refresh sync-skill .PHONY: help install test validate example diagram deck full clean lint codex-package update-icons freshness freshness-refresh sync-skill sku-discover
# Use venv if present, otherwise find best available python3 # Use venv if present, otherwise find best available python3
ifneq (,$(wildcard .venv/bin/python)) ifneq (,$(wildcard .venv/bin/python))
@@ -97,3 +97,6 @@ kb-check: ## KB freshness JSON (used by skill welcome-flow pre-flight)
freshness-refresh: ## Run refresh tools for stale KB files that support automation freshness-refresh: ## Run refresh tools for stale KB files that support automation
@$(PYTHON) tools/kb_freshness.py --auto-refresh @$(PYTHON) tools/kb_freshness.py --auto-refresh
sku-discover: ## Report SKUs present in Oracle API but missing from oci-sku-catalog.yaml
@$(PYTHON) tools/refresh_sku_catalog.py --discover

View File

@@ -560,6 +560,18 @@ categories:
list_price_usd: 0.0807 list_price_usd: 0.0807
default_hours_units: 730 default_hours_units: 730
billing_type: hourly billing_type: hourly
- sku: B95714
product: Oracle Autonomous AI Transaction Processing - Dedicated - ECPU
metric: ECPU Per Hour
list_price_usd: 0.336
default_hours_units: 730
billing_type: hourly
- sku: B95715
product: Oracle Autonomous AI Transaction Processing - Dedicated - ECPU - BYOL
metric: ECPU Per Hour
list_price_usd: 0.0807
default_hours_units: 730
billing_type: hourly
exascale: exascale:
display_name: Oracle Cloud Infrastructure - Database Exadata Exascale Infrastructure display_name: Oracle Cloud Infrastructure - Database Exadata Exascale Infrastructure
skus: skus:

View File

@@ -1,8 +1,11 @@
--- ---
last_verified: 2026-03-14 last_verified: 2026-04-24
service: OCI Compute service: OCI Compute
category: compute category: compute
oci_color: "#4F7B6E" oci_color: "#4F7B6E"
changelog:
- date: "2026-04-24"
change: "Added missing shapes surfaced by discover+WebFetch: VM/BM.Standard.E6.Ax, VM/BM.Standard.A4.Ax, and full X12 family (VM.Standard4.Ax.Flex 39 OCPU / 360 GB, BM.Standard4.Ax.120 latest-gen bare metal). Verified against docs.oracle.com/en-us/iaas/Content/Compute/References/computeshapes.htm."
--- ---
what: Virtual machines and bare metal instances. Flexible shapes allow custom what: Virtual machines and bare metal instances. Flexible shapes allow custom
@@ -42,6 +45,24 @@ shape_families:
max_ocpu: 18 max_ocpu: 18
memory_per_ocpu_gb: "1-64 GB per OCPU" memory_per_ocpu_gb: "1-64 GB per OCPU"
use_for: "High-frequency trading, HPC" use_for: "High-frequency trading, HPC"
- name: VM.Standard.E6.Ax.Flex
processor: "AMD EPYC (Ampere-branded E6 variant)"
max_ocpu: 94
max_memory_gb: 712
network_gbps: 99
use_for: "E6 Ampere-variant general purpose"
- name: VM.Standard.A4.Ax.Flex
processor: "Arm (A4 Ampere-branded variant)"
max_ocpu: 45
max_memory_gb: 720
network_gbps: 100
use_for: "A4 Ampere-variant Arm workloads"
- name: VM.Standard4.Ax.Flex
processor: "AMD EPYC (X12 Ampere-branded)"
max_ocpu: 39
max_memory_gb: 360
network_gbps: 99
use_for: "X12-generation general purpose (latest)"
- name: BM.Standard.E5 - name: BM.Standard.E5
processor: "AMD EPYC 9J14 (Genoa)" processor: "AMD EPYC 9J14 (Genoa)"
ocpus: 192 ocpus: 192
@@ -50,6 +71,30 @@ shape_families:
processor: "AMD EPYC 9J45 (Turin)" processor: "AMD EPYC 9J45 (Turin)"
ocpus: 256 ocpus: 256
use_for: "Bare metal latest generation" use_for: "Bare metal latest generation"
- name: BM.Standard.E6.Ax.192
processor: "AMD EPYC (E6 Ampere-branded variant)"
ocpus: 192
max_memory_gb: 1536
network_gbps: 200
use_for: "Bare metal E6 Ampere-variant"
- name: BM.Standard.A4.48
processor: "Arm (A4)"
ocpus: 48
max_memory_gb: 768
network_gbps: 100
use_for: "Bare metal A4 Arm general purpose"
- name: BM.Standard.A4.Ax.48
processor: "Arm (A4 Ampere-branded variant)"
ocpus: 48
max_memory_gb: 768
network_gbps: 100
use_for: "Bare metal A4 Ampere-variant Arm"
- name: BM.Standard4.Ax.120
processor: "AMD EPYC (X12 Ampere-branded)"
ocpus: 120
max_memory_gb: 1152
network_gbps: 200
use_for: "Bare metal X12 latest generation"
- name: VM.GPU.A10.1 - name: VM.GPU.A10.1
processor: "NVIDIA A10" processor: "NVIDIA A10"
gpu_count: 1 gpu_count: 1

View File

@@ -48,3 +48,4 @@ bom:
notes: notes:
- "Pricing based on Ashburn region (OC1)" - "Pricing based on Ashburn region (OC1)"
- "Discounts subject to Oracle approval" - "Discounts subject to Oracle approval"
- "Hours convention — 730 vs 744: this BOM uses 730 hrs/month (= 8,760 annual hrs / 12), which equals the actual yearly OCI billing for hourly SKUs. OCI Cost Estimator defaults to 744 hrs/month (worst-case 31-day month); extrapolated to 12 months, 744 overstates the real annual cost by 1.92%. For 12-month TCO/proposals, 730 is the accurate figure; 744 is only appropriate when quoting a single peak month as an upper bound."

View File

@@ -328,6 +328,7 @@ class OCIBomGenerator:
data_start_row = row data_start_row = row
cat_subtotal_rows = [] cat_subtotal_rows = []
data_item_rows = [] # line-item rows (excludes category headers/subtotals/totals)
# Ensure categories present in items but missing from catalog order # Ensure categories present in items but missing from catalog order
# (e.g., "other" from unknown SKUs) still render at the end. # (e.g., "other" from unknown SKUs) still render at the end.
@@ -418,6 +419,7 @@ class OCIBomGenerator:
ws.cell(row=row, column=COL_CONV_WO).number_format = '#,##0.00' ws.cell(row=row, column=COL_CONV_WO).number_format = '#,##0.00'
ws.cell(row=row, column=COL_CONV_W).number_format = '#,##0.00' ws.cell(row=row, column=COL_CONV_W).number_format = '#,##0.00'
data_item_rows.append(row)
row += 1 row += 1
# Category subtotal row # Category subtotal row
@@ -492,17 +494,15 @@ class OCIBomGenerator:
# ── Cost proportion formulas (back-fill) ───────────────── # ── Cost proportion formulas (back-fill) ─────────────────
# Cost % = line monthly w/ discount / total monthly w/ discount # Cost % = line monthly w/ discount / total monthly w/ discount
for item_row in range(data_start_row, total_row): mw_l = get_column_letter(COL_M_W)
cell_val = ws.cell(row=item_row, column=COL_M_W).value for item_row in data_item_rows:
if cell_val and str(cell_val).startswith("=") and "SUM" not in str(cell_val): ws.cell(
mw_l = get_column_letter(COL_M_W) row=item_row, column=COL_PCT,
ws.cell( value=f"=IF({mw_l}{total_row}=0,0,{mw_l}{item_row}/{mw_l}{total_row})",
row=item_row, column=COL_PCT, )
value=f"=IF({mw_l}{total_row}=0,0,{mw_l}{item_row}/{mw_l}{total_row})", ws.cell(row=item_row, column=COL_PCT).number_format = '0.0%'
) ws.cell(row=item_row, column=COL_PCT).font = data_font
ws.cell(row=item_row, column=COL_PCT).number_format = '0.0%' ws.cell(row=item_row, column=COL_PCT).border = thin_border
ws.cell(row=item_row, column=COL_PCT).font = data_font
ws.cell(row=item_row, column=COL_PCT).border = thin_border
# ── Notes ───────────────────────────────────────────────── # ── Notes ─────────────────────────────────────────────────
if self.notes: if self.notes:

View File

@@ -220,6 +220,13 @@ def refresh_catalog(verbose=False):
print(" Not in API (possibly retired): {}".format(not_found)) print(" Not in API (possibly retired): {}".format(not_found))
print(" Catalog saved: {}".format(CATALOG_PATH)) print(" Catalog saved: {}".format(CATALOG_PATH))
# Reverse-direction check: SKUs in API we haven't catalogued yet.
catalog_skus_after = {str(e["sku"]) for cat in raw.get("categories", {}).values()
for e in cat.get("skus", [])}
stub_map = {s: {} for s in catalog_skus_after}
new_skus = discover_missing_skus(api_map, stub_map, verbose=verbose)
print_missing_skus(new_skus, limit=None if verbose else 20)
def validate_catalog(verbose=False): def validate_catalog(verbose=False):
"""Compare current catalog prices against API and report differences.""" """Compare current catalog prices against API and report differences."""
@@ -282,6 +289,104 @@ def validate_catalog(verbose=False):
if len(missing) > 10: if len(missing) > 10:
print(" ... and {} more".format(len(missing) - 10)) print(" ... and {} more".format(len(missing) - 10))
# Reverse-direction check: SKUs in API we haven't catalogued yet.
new_skus = discover_missing_skus(api_map, sku_map, verbose=verbose)
print_missing_skus(new_skus, limit=None if verbose else 20)
def discover_missing_skus(api_map, catalog_sku_map, verbose=False):
"""Find SKUs present in the API but missing from the catalog.
The filter is auto-derived: we collect every `serviceCategory` value for
SKUs already in the catalog and only report new SKUs whose serviceCategory
falls within that curated set. This keeps the report relevant — as new
categories are added to the catalog, the discovery scope expands.
"""
curated_categories = set()
for sku in catalog_sku_map:
if sku in api_map:
sc = api_map[sku].get("serviceCategory", "") or ""
if sc:
curated_categories.add(sc)
if verbose:
print(" Curated serviceCategory values ({}): {}".format(
len(curated_categories), ", ".join(sorted(curated_categories))
))
missing = []
for sku, product in api_map.items():
if sku in catalog_sku_map:
continue
sc = product.get("serviceCategory", "") or ""
if sc not in curated_categories:
continue
missing.append({
"sku": sku,
"product": product.get("displayName", ""),
"service_category": sc,
"metric": product.get("metricName", ""),
"price": extract_payg_price(product),
})
missing.sort(key=lambda x: (x["service_category"], x["sku"]))
return missing
def print_missing_skus(missing, limit=None):
"""Print a readable report of missing SKUs, grouped by serviceCategory."""
if not missing:
print("\nNo new SKUs discovered in curated categories. Catalog covers the API well.")
return
print("\n{} SKUs present in API but missing from catalog".format(len(missing)))
print("(filtered to serviceCategory values already represented in our catalog)\n")
by_cat = {}
for m in missing:
by_cat.setdefault(m["service_category"], []).append(m)
shown = 0
truncated = False
for cat, items in sorted(by_cat.items()):
print(" [{}] ({} new)".format(cat, len(items)))
for m in items:
if limit is not None and shown >= limit:
truncated = True
break
price_str = "${:.4f}".format(m["price"]) if m["price"] else " -"
print(" {:<10} {:>10} {}".format(
m["sku"], price_str, m["product"][:70]
))
shown += 1
if truncated:
break
if truncated:
print("\n ... and {} more (re-run with -v to list all)".format(len(missing) - shown))
print("\nNext step: review and add relevant entries to kb/pricing/oci-sku-catalog.yaml")
print(" under the appropriate category block.")
def discover_catalog(verbose=False):
"""Stand-alone --discover entry point: report new API SKUs not yet in catalog."""
print("Fetching current products from Oracle API...")
products = fetch_all_products(currency=DEFAULT_CURRENCY, verbose=verbose)
api_map = {}
for p in products:
pn = p.get("partNumber", "")
if pn:
api_map[pn] = p
raw, sku_map = load_current_catalog()
if not raw:
print("Error: Could not load catalog at {}".format(CATALOG_PATH))
return 1
missing = discover_missing_skus(api_map, sku_map, verbose=verbose)
print_missing_skus(missing, limit=None if verbose else 40)
return 0
def inspect_sku(part_number): def inspect_sku(part_number):
"""Fetch and display a single SKU from the API.""" """Fetch and display a single SKU from the API."""
@@ -571,6 +676,9 @@ def main():
"(currently: compute)") "(currently: compute)")
parser.add_argument("--validate", action="store_true", parser.add_argument("--validate", action="store_true",
help="Validate catalog prices against API") help="Validate catalog prices against API")
parser.add_argument("--discover", action="store_true",
help="Report SKUs present in API but missing from catalog "
"(filtered to serviceCategory values already curated)")
parser.add_argument("--sku", type=str, parser.add_argument("--sku", type=str,
help="Inspect a single SKU from the API") help="Inspect a single SKU from the API")
parser.add_argument("--dump", type=str, metavar="FILE", parser.add_argument("--dump", type=str, metavar="FILE",
@@ -589,6 +697,8 @@ def main():
refresh_catalog(verbose=args.verbose or args.diff) refresh_catalog(verbose=args.verbose or args.diff)
elif args.validate: elif args.validate:
validate_catalog(verbose=args.verbose) validate_catalog(verbose=args.verbose)
elif args.discover:
return discover_catalog(verbose=args.verbose)
elif args.dump: elif args.dump:
dump_all(args.dump, verbose=args.verbose) dump_all(args.dump, verbose=args.verbose)
else: else: