spb/datacenterindex

1id: cloudhq2name: CloudHQ3domain: cloudhq.com4kind: operator5mode: hybrid6priority: 27license: "Publicly accessible operator pages; factual data only; © CloudHQ"8attribution: "CloudHQ"9homepage: https://cloudhq.com/campuses/10notes: "The `campus` post type is absent from sitemaps.xml, so campus pages are discovered by following links from /campuses/ (3 pages). Campus pages /campus/<slug>/ publish a coordinate line (lat, lng), total critical IT load (MW), square feet and RFS timing (→ under_construction when RFS is in the future)."11coverage: { operators: ["CloudHQ"], countries: ["US", "GB", "NL", "IT", "DE", "FR", "JP", "TH", "BR", "MX", "PY"] }12fetch: { level: 1, maxLevel: 2, rpm: 15, concurrency: 2, respectRobots: true, maxCreditsPerRun: 20 }13discovery:14  sitemap: ["https://cloudhq.com/sitemaps.xml"]15  seeds: [{ url: "/campuses/", group: seed, pageType: facility_index }, { url: "/campuses/?sf_paged=2", group: seed, pageType: facility_index }, { url: "/campuses/?sf_paged=3", group: seed, pageType: facility_index }]16  follow: ["/campus/[a-z0-9-]+/$"]17  include: ["/campus/[a-z0-9-]+/$", "/news/", "/press/"]18  exclude: ["\\?", "#"]19  classify:20    - { pattern: "/campus/[a-z0-9-]+/$", group: facility_pages, pageType: facility_page, priority: 60 }21    - { pattern: "/(news|press)/.", group: newsroom, pageType: press_release, priority: 50 }22schedule: { facility_pages: weekly, newsroom: daily, seed: monthly, index: monthly, sitemap: weekly }23extractors:24  facility: { match: ["/campus/[a-z0-9-]+/$"], kind: facility, parser: cloudhq_campus_v1 }25defaults: { operatorName: CloudHQ, facilityType: hyperscale, isHyperscale: true }26