Intranet portals: SSO attachments plus CDN long-cache mean “I replaced the image but everyone still sees the old one”—surface version tokens in URLs or filenames
`intranet-visio-jpg` targets knowledge-base hero banners, change-announcement illustrations, or NOC wallboard assets. Corporate proxies often recompress images and caches linger, so staff still fetch stale JPGs after ops published a fix, triggering “did we actually ship it?” disputes. Bake a version fragment into each object key at storage or gateway layers, and avoid indefinite Cache-Control: immutable on assets unless HTML and paths bump together.
Intranet runbook: align portal theme colors and banner crop, purge-cache walkthrough on staging, then register object keys against each .vsdx revision
- Confirm banner safe zones, max upload size, and forced sRGB with the portal team; export JPGs at the exact banner pixel grid from .vsdx so front-end CSS scaling does not magnify JPEG blocking.
- Test pre-prod URLs in incognito and with a legacy employee profile; if stale objects persist, check the wrong CDN edge or a rogue Service Worker before blaming the converter.
- Change posts must cite attachment hashes or version ids—never “see attached figure” with no anchor; emergency rollback should swap object keys quickly and broadcast an explicit cache-bust directive.
VSDX to JPG (intranet) FAQ: cache vs object keys, banner crops, proxy compression, auditable announcements
Staff say the announcement graphic is still last week’s even though object storage updated—do we blame browser cache or reverse-proxy cache first?
Investigate both in parallel: verify HTML still references the old URL, whether the CDN needs purge, and whether mail clients inlined an offline copy; curl with Cache-Control: no-cache to compare edge vs origin bytes.
Compliance scanners flag EXIF camera metadata in JPGs—do schematic exports leak too, and how do we strip it?
Server-side exports should strip metadata by default or run an exiftool-style scrub in the release pipeline; even diagrams may carry generator fingerprints, so apply one security baseline for all JPGs.
The portal offers light and dark themes; a single white-matte JPG glares in dark mode—how do we avoid maintaining two hand-tuned Visio files forever?
Ship theme-specific backgrounds or prefer transparent PNG/WebP; if JPG is mandatory, bind different media assets to theme fields in the CMS instead of one file forced onto both skins.
Ops wallboards refresh every few seconds and full-page JPGs flicker—should we move to video, Canvas vectors, or JSON push?
Sub-second refresh with repeated full bitmap decode causes flash; evaluate browser Canvas drawing or WebSocket-fed topology JSON rather than re-downloading a giant JPG each tick.
Multi-region intranet mirrors lag so North China sees the new art while South still shows the old—how should releases bind a global object key to the announced effective time?
Wait for mirror consistency or reference a single global origin hostname; the announcement body should state go-live time plus checksums so staggered caches do not imply the change failed.