# ironmountain — Iron Mountain Data Centers (DISABLED)


Config	`config/connectors/ironmountain.yaml` (`enabled: false`)
Parser	none shipped
Kind / priority	operator / 1
License / attribution	Publicly accessible operator pages; factual data only; © Iron Mountain Incorporated

# Why it is disabled (2026-09-11)

Every direct request to www.ironmountain.com — robots.txt, /sitemap.xml, /data-centers, /data-centers/global-data-center-locations, with the bot UA and with a browser UA — returns HTTP 429 with a "Vercel Security Checkpoint" JavaScript challenge. This is a hard bot wall, not a rate limit we can wait out.
Scrapfly with ASP does pass (HTTP 200), but each request is billed 30 credits (residential proxy 25 + browser 5, even with render_js=false). Discovery alone needs robots.txt + the sitemap index + 9 sitemap shards (sitemap1.xml … sitemap9.xml) ≈ 330 credits before the first facility page, i.e. more than the whole 150-credit run budget, and every weekly recrawl would cost the same again. That violates the "premium only when blocked, never for bulk crawling" rule and the daily budget.
Two probe requests were spent to establish this (60 credits): robots.txt and the sitemap index. robots.txt (fetched through Scrapfly) allows /data-centers/, disallows Sitecore/system folders, /zh-cn/, ?localize= and some landing pages, and blocks the Scrapy and BoogleBot2 agents.

# What is in the YAML

Discovery/classify rules are drafted from the public URL scheme (/data-centers/…, newsroom) but no extractor is declared — the facility page template could not be inspected without spending credits, and shipping unverified extraction is not allowed. The connector is enabled: false so the scheduler ignores it.

# How to re-enable

Check whether the challenge has been lifted: curl -A "DataCenterIndexBot/0.1" https://www.ironmountain.com/robots.txt should return 200 text.
Inspect 2–3 facility pages, confirm the URL pattern (/data-centers/<region>/<facility> is an assumption) and write ironmountain_facility_v1 in apps/worker/src/connectors/operators1/.
Verify with try-connector (--limit 4, spot-check two sites), then set enabled: true. If the wall stays, an alternative is Iron Mountain's PeeringDB / registry presence via the registry connectors rather than the operator site.