Bulk migration: reconciled inventories, failure buckets, collision-proof names
`batch-vsdx-docx` moves whole libraries into Word for search/template programs. Risks include basename collisions, OLE-heavy files stalling queues, and partial failures masked as success. Pull manifests with paths and checksums, pilot to classify errors, then scale concurrency. Reconciliation reports must list success hashes, skips, and backlog—otherwise “we migrated the folder” is unprovable. Split heavy embed jobs from light flowcharts to protect SLAs.
Bulk VSDX→DOCX: path-keyed manifests, bucketed failures, hash reconciliation
- Export library indexes with relative paths, sizes, version ids, and owners; chunk by product or site with caps—no unlogged drag-and-drop of entire drives.
- Pilot-run to capture corrupt, encrypted, oversize, timeout, and dependency errors with playbooks each; lock output names/hashes for successes—renames need change control.
- Ship per-batch reconciliation and use checksum deltas to auto-queue reconversion, preventing silent stale PDF/DOCX pairs.
VSDX to DOCX (batch) FAQ
Same filename in different folders—overwrite risk?
Yes if outputs ignore paths; key on full relative paths or content hashes and surface collisions for renames before enqueue.
Mixed success/failure—how to message consumers?
Freeze successful subset hashes per batch and publish a separate failure backlog—do not claim the whole library is done.
Timeouts—retry forever?
Use capped backoff and error classes; infinite retries hide root causes and burn capacity—keep first-failure diagnostics.
Search still misses device IDs after migration—why?
IDs may be inside images or alt text only—promote them to searchable paragraphs/tables in Visio before reconverting.
Owner transition—minimum handoff?
Parameter versions, golden files, reconciliation scripts, skip dictionaries, and open backlog—without that, audits fail months later.