Glacier.sync()

sync() reconciles cache state across edge locations so warmed and invalidated entries stay consistent no matter which region a user lands in.

Signature

glacier.sync(options?: SyncOptions): Promise<SyncResult>

Why you need it

In a single-region deployment, the cache is one thing in one place and always consistent. Across multiple edge regions, each node keeps its own warm set. Without coordination, a user who moves between regions — or whose request is routed to a different node — can hit a cold or stale cache.

sync() exchanges a compact digest of cache state between nodes and reconciles the differences: warming entries that are hot elsewhere, and dropping ones invalidated in another region.

Options

• regions — limit the sync to specific regions instead of all of them.
• mode — 'push' shares this node's state, 'pull' adopts others', 'merge' (default) does both.

Example

Most apps never call sync() directly — Glacier runs it on an interval automatically. You'd call it manually after a large batch change that you want reflected everywhere immediately:

import { glacier } from '../glacier.config'

await glacier.invalidate('catalog')
const result = await glacier.sync({ mode: 'push' })

console.log(`Synced ${result.regions.length} regions`)

Return value

Resolves with a SyncResult describing which regions were reached and how many entries were warmed or dropped. See Observability for how to monitor sync activity in production.