autobee 2.0.0-rc.23 → 2.0.0-rc.26

package/index.js

@@ -14,6 +14,7 @@ const asserts = require('./lib/asserts.js')
 const boot = require('./lib/boot.js')
 const { resolveWeight, currentWeight } = require('./lib/witness.js')
 const encoding = require('./lib/encoding.js')
+const { LEGACY_OPLOG_VERSION } = require('./lib/constants.js')
 const FastForward = require('./lib/fast-forward.js')
 const System = require('./lib/system.js')
 const ApplyCalls = require('./lib/apply-calls.js')
@@ -22,7 +23,7 @@ const { ActiveWriters } = require('./lib/writers.js')
 const TrustedPeers = require('./lib/trusted.js')
 const ApplyView = require('./lib/apply-view.js')
 const UpdateChanges = require('./lib/updates.js')
-const migrations = require('./lib/migrations')
+const migrations = require('./lib/migrations.js')
 
 const EMPTY_HEAD = { length: 0, key: null }
 const INTERRUPT = new Error('Apply interrupted')
@@ -527,12 +528,12 @@ module.exports = class Autobee extends ReadyResource {
     this.stats.drains++
 
     if (this.bootFrom) {
-      const { head = null, legacy = null, bootCondition = null } = this.bootFrom
+      const { head = null, legacy = null, bootCondition = null, wait = false } = this.bootFrom
 
       this.bootFrom = null
 
       if (legacy) await this._bootFromSystem(legacy)
-      else if (head) await this._bootFromHead(head, bootCondition)
+      else if (head) await this._bootFromHead(head, bootCondition, wait)
     }
 
     this._ackRequired = false
@@ -1195,9 +1196,9 @@ module.exports = class Autobee extends ReadyResource {
 
   // the boot head seeds the view every trust decision here is made against,
   // since our own view is still empty
-  async _bootFromHead(head, bootCondition) {
+  async _bootFromHead(head, bootCondition, wait = false) {
     // just a head: one attempt, we do not wait around if it cannot be read
-    if (bootCondition === null) {
+    if (bootCondition === null && !wait) {
       try {
         const ff = await FastForward.fromHead(this, head, null, {
           force: true,
@@ -1211,8 +1212,8 @@ module.exports = class Autobee extends ReadyResource {
       }
     }
 
-    // a condition needs the view at head to check against, so park until we
-    // can read it and something acceptable turns up
+    // park until the head can be read and, if a condition is set, something
+    // acceptable turns up
     let opened = null
 
     try {
@@ -1220,10 +1221,28 @@ module.exports = class Autobee extends ReadyResource {
         this._bootWait = rrp()
 
         try {
-          if (opened === null) opened = await this._bootReference(head)
-
-          if (opened !== null && (await this._bootAttempt(head, bootCondition, opened.view))) {
-            return true
+          if (bootCondition === null) {
+            const ff = await FastForward.fromHead(this, head, null, {
+              force: true,
+              timeout: FastForward.DEFAULT_TIMEOUT
+            })
+
+            if (ff !== null && (await this._runFastForward(ff))) return true
+          } else {
+            if (opened === null) opened = await this._bootReference(head)
+
+            if (opened !== null) {
+              if (await this._bootAttempt(head, bootCondition, opened.view)) {
+                return true
+              }
+
+              // the reference may be stale - resolve it afresh next attempt
+              await opened.close()
+              opened = null
+            } else if (await this._bootLegacyHead(head)) {
+              // boot legacy head through the migration pathway
+              return true
+            }
           }
         } catch (err) {
           safetyCatch(err)
@@ -1258,6 +1277,20 @@ module.exports = class Autobee extends ReadyResource {
     return ff !== null && (await this._runFastForward(ff))
   }
 
+  async _bootLegacyHead(head) {
+    const oplog = await FastForward.flushHead(this, head, {
+      timeout: FastForward.DEFAULT_TIMEOUT
+    })
+    if (oplog === null || oplog.op.version > LEGACY_OPLOG_VERSION) return false
+
+    const ff = await FastForward.fromHead(this, head, null, {
+      force: true,
+      timeout: FastForward.DEFAULT_TIMEOUT
+    })
+
+    return ff !== null && (await this._runFastForward(ff))
+  }
+
   async _bootReference(head) {
     const oplog = await FastForward.flushHead(this, head, {
       timeout: FastForward.DEFAULT_TIMEOUT

package/lib/boot.js

@@ -1,5 +1,5 @@
 const encoding = require('./encoding.js')
-const { checkAutobaseMigration } = require('./migrations')
+const { checkAutobaseMigration } = require('./migrations.js')
 
 module.exports = async function boot(
   corestore,

package/lib/fast-forward.js

@@ -3,7 +3,7 @@ const b4a = require('b4a')
 const Hypercore = require('hypercore')
 
 const System = require('./system.js')
-const migrations = require('./migrations')
+const migrations = require('./migrations.js')
 const { LEGACY_AUTOBASE_VERSION } = require('./constants.js')
 
 const DEFAULT_OP_TIMEOUT = 5_000

package/lib/migrations.js

@@ -0,0 +1,421 @@
+const c = require('compact-encoding')
+const b4a = require('b4a')
+const crypto = require('hypercore-crypto')
+const { AutobeeEncryption } = require('autobee-encryption')
+const { decodeBlock } = require('hyperbee2/lib/encoding.js')
+
+const encoding = require('./encoding.js')
+const { assert, bail } = require('./asserts.js')
+const { LEGACY_OPLOG_VERSION } = require('./constants.js')
+
+const { getEncoding } = require('../encoding/spec/autobee')
+
+const SystemInfoV1 = getEncoding('@autobase-compat/info-v1')
+const SystemInfoV2 = getEncoding('@autobase-compat/info-v2')
+
+const EMPTY = b4a.alloc(0)
+const INDEX_VERSION = 1
+const [NS_SIGNER_NAMESPACE] = crypto.namespace('autobase', 1)
+
+// indexer rotations chain legacy systems - how many generations we chase
+const MAX_MIGRATE_HOPS = 64
+
+module.exports = {
+  MAX_MIGRATE_HOPS,
+  checkAutobaseMigration,
+  readLegacySystemInfo,
+  findNewestSystem,
+  resolveLegacyViews,
+  inflateLegacyOplog,
+  decodeLegacySystemInfo,
+  deriveNamespace,
+  coreLength
+}
+
+// boot-time migration of local legacy autobase storage
+async function checkAutobaseMigration(store, local, bootstrap, legacyViews) {
+  const bootRecord = await local.getUserData('autobase/boot')
+  if (!bootRecord) return null
+
+  const { key } = encoding.decodeAutobaseBootRecord(bootRecord)
+  const core = store.get({ key: key, encryption: null })
+  await core.ready()
+
+  const system = { key, length: core.length }
+
+  // setup encryption
+  const encryptionKey = await local.getUserData('autobase/encryption')
+  await AutobeeEncryption.setSystemEncryption(bootstrap.key, encryptionKey, core)
+
+  const catchup = await getCatchupHeads(core)
+  const nodes = await Promise.all(
+    catchup.map((n) => getWriterBatch(store, n, bootstrap.key, encryptionKey))
+  )
+
+  const info = await readLegacySystemInfo(core, core.length)
+
+  const indexerManifests = await Promise.all(
+    info.indexers.map((idx) => storeCoreManifest(store, idx.key))
+  )
+  const views = await Promise.all(info.views.map((v) => getPersistedView(store, v)))
+  const entropy = legacyEntropy(info, indexerManifests)
+
+  const getManifest = (key) => storeCoreManifest(store, key)
+
+  const mappedViews = new Map()
+  for (const name of legacyViews) {
+    const v = await findViewByName(
+      getManifest,
+      bootstrap.key,
+      encryptionKey,
+      indexerManifests,
+      views,
+      entropy,
+      name
+    )
+
+    mappedViews.set(name, v)
+  }
+
+  await core.close()
+
+  return {
+    encryptionKey,
+    system,
+    views: mappedViews,
+    catchup: nodes
+  }
+}
+
+// decode the legacy system info recorded in the system core's block at length
+async function readLegacySystemInfo(core, length, opts = null) {
+  // Decode hyperbee block using hyperbee2 compat
+  const node = decodeBlock(await core.get(length - 1, opts))
+  assert(node.keys.length > 0, 'bad system block')
+
+  // Decode system info from hyperbee block using autobee compat
+  return decodeLegacySystemInfo(node.keys[0].value)
+}
+
+// read each indexer's newest legacy oplog stamp - its digest/checkpoint mark
+// the newest legacy system that indexer knows. take the majority vote. an
+// indexer that already moved to autobee announces its live autobee state
+// instead - keep the newest of those as a direct fast-forward candidate and
+// bisect down to its buried legacy stamps so its vote still counts
+async function findNewestSystem(ff, info) {
+  const tally = new Map()
+
+  let autobee = null
+
+  for (const idx of info.indexers) {
+    const core = ff.auto.openCore(idx.key)
+    ff.cores.push(core)
+    await core.ready()
+
+    const length = await coreLength(core, ff.timeout)
+    if (!length || ff.destroyed) continue
+
+    let oplog = await flushHead(ff.auto, { key: idx.key, length }, { timeout: ff.timeout })
+    if (oplog === null || !oplog.op.views) continue
+
+    if (oplog.op.version > LEGACY_OPLOG_VERSION) {
+      if (autobee === null || oplog.op.views.flushes > autobee.op.views.flushes) {
+        autobee = oplog
+      }
+
+      const boundary = await findLegacyBoundary(ff, core, idx.length, oplog.length - 1)
+      if (!boundary || ff.destroyed) continue
+
+      oplog = await flushHead(ff.auto, { key: idx.key, length: boundary }, { timeout: ff.timeout })
+      if (oplog === null || !oplog.op.views) continue
+      if (oplog.op.version > LEGACY_OPLOG_VERSION) continue
+    }
+
+    const sys = oplog.op.views.system
+    if (!sys || !sys.key) continue
+
+    const hex = b4a.toString(sys.key, 'hex')
+    const t = tally.get(hex) || { key: sys.key, votes: 0, length: sys.length }
+
+    t.votes++
+    if (sys.length > t.length) t.length = sys.length
+    tally.set(hex, t)
+  }
+
+  const majority = (info.indexers.length >> 1) + 1
+
+  let best = null
+  for (const t of tally.values()) {
+    if (t.votes >= majority && (best === null || t.length > best.length)) best = t
+  }
+
+  return { legacy: best === null ? null : best.key, autobee }
+}
+
+// map the legacy views recorded in the system info by name
+async function resolveLegacyViews(ff, info) {
+  const indexerManifests = await Promise.all(
+    info.indexers.map((idx) => ffCoreManifest(ff, idx.key, idx.length))
+  )
+  if (ff.destroyed || indexerManifests.some((m) => m === null)) return null
+
+  const entropy = legacyEntropy(info, indexerManifests)
+
+  const getManifest = (key, length) => ffCoreManifest(ff, key, length)
+
+  const views = new Map()
+
+  await Promise.all(
+    ff.auto.legacyViews.map(async (name) => {
+      const v = await findViewByName(
+        getManifest,
+        ff.auto.key,
+        ff.auto.encryptionKey,
+        indexerManifests,
+        info.views,
+        entropy,
+        name
+      )
+
+      if (v) views.set(name, v)
+    })
+  )
+
+  if (ff.destroyed) return null
+
+  return views
+}
+
+// legacy nodes always inflate, but only indexers carry the digest/checkpoint
+// stamps that synthesise system info - callers must check op.views
+async function inflateLegacyOplog(buf, core, seq, opts = null) {
+  const m = encoding.decodeRawOplog(buf)
+
+  const op = {
+    version: m.version,
+    timestamp: 0,
+    links: m.node.heads,
+    batch: { start: 0, end: m.node.batch - 1 },
+    views: null,
+    optimistic: !!m.optimistic,
+    value: m.node.value
+  }
+
+  if (m.digest === null || m.checkpoint === null || !m.checkpoint.system) return op
+
+  const fetches = []
+
+  fetches.push(m.digest.pointer ? core.get(seq - m.digest.pointer, opts) : buf)
+  fetches.push(m.checkpoint.pointer ? core.get(seq - m.checkpoint.system.checkpointer, opts) : buf)
+
+  const [digestNode, checkpointNode] = await Promise.all(fetches)
+  // a wait: false get we could not satisfy locally
+  if (digestNode === null || checkpointNode === null) return op
+
+  const { digest } = encoding.decodeRawOplog(digestNode)
+  const { checkpoint } = encoding.decodeRawOplog(checkpointNode)
+
+  if (!checkpoint.system || !checkpoint.system.checkpoint) return op
+
+  op.views = {
+    system: {
+      key: digest.key,
+      start: 0,
+      length: checkpoint.system.checkpoint.length
+    },
+    flushes: seq
+  }
+
+  return op
+}
+
+function decodeLegacySystemInfo(buffer) {
+  const state = { start: 0, end: buffer.length, buffer }
+  const version = c.uint.decode(state)
+
+  state.start--
+
+  switch (version) {
+    case 1:
+      return SystemInfoV1.decode(state)
+    case 2:
+      return SystemInfoV2.decode(state)
+    default:
+      bail('Expected legacy system info')
+  }
+}
+
+function deriveNamespace(name, bootstrap, entropy, encryptionKey) {
+  const encryptionId = crypto.hash(encryptionKey || EMPTY)
+  const version = c.encode(c.uint, INDEX_VERSION)
+
+  return crypto.hash([
+    NS_SIGNER_NAMESPACE,
+    version,
+    bootstrap,
+    encryptionId,
+    entropy,
+    b4a.from(name)
+  ])
+}
+
+function coreLength(core, timeout) {
+  if (core.length) return core.length
+
+  return new Promise((resolve) => {
+    core.on('append', () => resolve(core.length))
+    setTimeout(resolve, timeout, 0)
+  })
+}
+
+// a writer's oplog is a legacy prefix followed by autobee-format nodes -
+// bisect for the boundary. lo is a length known to be legacy (the legacy
+// system's recorded length for this writer), hi a seq known to be autobee.
+// returns the legacy prefix length
+async function findLegacyBoundary(ff, core, lo, hi) {
+  while (lo < hi) {
+    const mid = (lo + hi) >>> 1
+
+    const block = await core.get(mid, {
+      timeout: ff.timeout,
+      activeRequests: ff.activeRequests
+    })
+
+    if (encoding.decodeRawOplog(block).version > LEGACY_OPLOG_VERSION) hi = mid
+    else lo = mid + 1
+  }
+
+  return lo
+}
+
+function legacyEntropy(info, indexerManifests) {
+  return info.version > 1 && info.entropy ? info.entropy : indexerManifests[0].signers[0].namespace
+}
+
+async function findViewByName(
+  getManifest,
+  bootstrap,
+  encryptionKey,
+  indexerManifests,
+  views,
+  entropy,
+  name
+) {
+  if (indexerManifests.length === 0) return null
+
+  const namespace = deriveNamespace(name, bootstrap, entropy, encryptionKey)
+
+  for (const v of views) {
+    const manifest = await getManifest(v.key, v.length)
+    if (!manifest) continue
+
+    if (manifest.signers.length === 0) continue
+
+    const signer = manifest.signers[0]
+
+    if (b4a.equals(signer.namespace, namespace)) return v
+  }
+
+  return null
+}
+
+// request-tracked manifest read over replication
+async function ffCoreManifest(ff, key, length) {
+  const core = ff.auto.store.get(key)
+  ff.cores.push(core)
+
+  await core.ready()
+
+  try {
+    if (!core.manifest) {
+      await core.get(length - 1, {
+        timeout: ff.timeout,
+        activeRequests: ff.activeRequests
+      })
+    }
+  } catch {
+    return null
+  }
+
+  return core.manifest
+}
+
+// local-storage manifest read
+async function storeCoreManifest(store, key) {
+  const core = store.get(key)
+  await core.ready()
+
+  const manifest = core.manifest
+  await core.close()
+
+  return manifest
+}
+
+async function getPersistedView(store, view) {
+  const core = store.get(view.key)
+  await core.ready()
+
+  const length = core.length
+  await core.close()
+
+  return { key: view.key, length }
+}
+
+async function getCatchupHeads(core) {
+  const session = core.session({ name: 'batch' })
+  await session.ready()
+
+  const nodes = []
+  for (let i = core.length; i < session.length; i++) {
+    const node = await session.get(i, { wait: false })
+    if (!node) throw new Error('Expect nodes to exist locally')
+
+    const entry = decodeBlock(node).keys[0]
+    if (entry.key[0] !== 0x01) continue
+
+    const { key, length } = encoding.decodeSystemWriter(entry.key, entry.value)
+    if (!length) continue
+    nodes.push({ key, length })
+  }
+
+  await session.close()
+
+  return nodes
+}
+
+async function getWriterBatch(store, head, key, encryptionKey, nodes = []) {
+  const batch = []
+
+  const core = store.get(head.key)
+  await core.ready()
+
+  if (encryptionKey) {
+    await core.setEncryption(AutobeeEncryption.getWriterEncryption(key, encryptionKey))
+  }
+
+  let seq = head.length - 1
+  const block = await core.get(seq--, { wait: false })
+  if (!block) throw new Error('Expect writer node to exist locally')
+
+  const node = encoding.decodeOplog(block)
+
+  batch.unshift({ ...head, ...node, from: core })
+
+  while (seq >= 0) {
+    const block = await core.get(seq--, { wait: false })
+    if (!block) break
+
+    const node = encoding.decodeOplog(block)
+    if (!node.batch.end) break
+
+    batch.unshift({ key: head.key, length: seq + 2, from: core, ...node })
+  }
+
+  return batch
+}
+
+// lazy to avoid the require cycle - fast-forward imports this module
+function flushHead(auto, head, opts) {
+  const FastForward = require('./fast-forward.js')
+  return FastForward.flushHead(auto, head, opts)
+}

package/lib/writers.js

@@ -59,21 +59,26 @@ class ActiveWriters {
     let w = this.active.get(id)
 
     if (w === this.localWriter) return
+
+    let added = false
     if (!w) {
       // anchors are local-only and never replicated - nothing to fetch
       const info = await this.auto.system.get(key, { unflushed: true })
       if (info && info.isAnchor) return
       w = await this.add(key)
       if (!w) return
+      added = true
     }
 
+    const advanced = length > w.hintedLength
+
     // remember the announced length so gc doesn't drop a writer we haven't caught up on
     w.hint(length)
 
     // prefetch the oplog head for the fast-forward that runs right after us
     if (w && length > 0) w.core.get(length - 1).catch(noop)
 
-    this.bump()
+    if (added || advanced) this.bump()
   }
 
   getLatestLocalOplog() {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "autobee",
-  "version": "2.0.0-rc.23",
+  "version": "2.0.0-rc.26",
   "description": "Bee based autobase",
   "license": "Apache-2.0",
   "author": "Holepunch Inc.",
@@ -43,6 +43,7 @@
     "lunte": "^1.6.0",
     "prettier": "^3.8.1",
     "prettier-config-holepunch": "^2.0.0",
-    "protomux-wakeup": "^2.9.0"
+    "protomux-wakeup": "^2.9.0",
+    "replication-simulator": "^3.0.0"
   }
 }