Add a bunch of debug statements useful when debugging sync issues

index.js

@@ -17,4 +17,8 @@ async function main () {
   })
 }
 
+process.on('unhandledRejection', error => {
+  console.error('==> Got unhandled rejection:', error.message)
+})
+
 main().catch(e => console.error(`top-level exception: ${e} json: ${JSON.stringify(e)}`))

lib/api.js

@@ -3,6 +3,7 @@ import Debug from 'debug'
 import Koa from 'koa'
 import Router from '@koa/router'
 import { addDebugRoutes } from './debug-routes.js'
+import DistributedStorage from './distributed-storage.js'
 
 const debug = Debug('whalesong:api')
 
@@ -111,8 +112,11 @@ export async function setupApp (storage, host, port, externalUrl) {
 
   router.head('/v2/:org/:name/manifests/:tag', async (ctx) => {
     const { org, name, tag } = ctx.params
+    debug(`Client requested (HEAD) (org,name,tag): (${org}, ${name}, ${tag})`)
     const pubKey = await lookupOrg(ctx, org)
+    debug(`Found pubkey ${pubKey} from org ${org}`)
     const { digest, size, contentType } = await storage.hasManifest(pubKey, name, tag)
+    debug(`hasManifest returned digest and size ${digest} size ${size}`)
     if (digest != null) {
       ctx.body = null
       ctx.set('Docker-Content-Digest', digest)
@@ -127,8 +131,11 @@ export async function setupApp (storage, host, port, externalUrl) {
 
   router.get('/v2/:org/:name/manifests/:tag', async (ctx) => {
     const { org, name, tag } = ctx.params
+    debug(`Client requested (GET) (org,name,tag): (${org}, ${name}, ${tag})`)
     const pubKey = await lookupOrg(ctx, org)
+    debug(`Found pubkey ${pubKey} from org ${org}`)
     const { digest, stream, contentType } = await storage.getManifest(pubKey, name, tag)
+    debug(`getManifest returned digest ${digest} stream ${stream}`)
     if (digest != null) {
       console.debug(`Retrieving manifest with digest ${digest}`)
       ctx.body = stream

lib/distributed-clients.js

@@ -95,8 +95,8 @@ async function setupHyperbeeClient (feed, pubKey, shouldReplicate) {
   const corePath = path.join(await baseDir(), 'hypercores', feed)
 
   // TODO: to sparse, or not to sparse, that is the question.
-  const core = hypercore(corePath, pubKey)
-  // const core = hypercore(corePath, pubKey, { sparse: true })
+  // const core = hypercore(corePath, pubKey)
+  const core = hypercore(corePath, pubKey, { sparse: true })
 
   const bee = new Hyperbee(core, {
     keyEncoding: 'utf-8',

lib/distributed-storage.js

@@ -33,11 +33,13 @@ class DistributedStorage {
   blobStore (org) {
     // returns the blob sub-database for the hyperbee for a given organization if it exists, otherwise nullish
     const bee = this.hyperbees.get(org)
+    debug(`returning blob store for org ${org}, got bee ${bee}}`)
     return bee?.sub('blob')
   }
 
   manifestStore (org, name) {
     // returns the manifest sub-database for the hyperbee for a given organization if it exists, otherwise nullish
+    debug(`returning manifest store for org ${org}`)
     const bee = this.hyperbees.get(org)
     return bee?.sub('manifest').sub(name)
   }
@@ -46,16 +48,19 @@ class DistributedStorage {
     const blobs = this.blobStore(org)
     // TODO: should we make an initial sync here too just like for manifests if org. does not exist?
     if (blobs) {
+      debug(`getting blob ${dockerDigest}`)
       const blobObjectNode = await blobs.get(dockerDigest)
       const blobObject = blobObjectNode?.value
 
+      debug(`blob lookup resulted in ${JSON.stringify(blobObject)}`)
       if (!blobObject) {
         return { stream: null, size: null, contentType: null }
       }
 
       const { contentType, size } = blobObject
       const ipfsCid = blobObject.locations[0].ref // currently, support only ipfs cid and a single location for every blob
 
+      debug(`blob lookup yielded ipfs cid ${ipfsCid}`)
       if (!ipfsCid) {
         return { stream: null, size: null, contentType: null }
       }
@@ -102,7 +107,9 @@ class DistributedStorage {
 
   async hasBlob (org, name, digest) {
     const blobs = this.blobStore(org)
+    debug(`trying to check if blob ${digest} exists`)
     const blobObjectNode = await blobs?.get(digest)
+    debug(`blob object node is ${JSON.stringify(blobObjectNode)}`)
     if (blobObjectNode) {
       const { contentType, size } = blobObjectNode?.value
       return { contentType, size }
@@ -177,9 +184,11 @@ class DistributedStorage {
       debug(`got manifest for ${org}/${name}:${tagOrDigest} returned digest ${manifestDigest}`)
     }
 
+    debug(`trying to get manifestDigest ${manifestDigest}`)
     if (manifestDigest) {
       // We now know the digest to use. Grab it!
       const { stream, size, contentType } = await this.getBlob(org, name, manifestDigest)
+      debug(`got blob with size ${size}`)
       if (stream != null) {
         return { digest: manifestDigest, stream, size, contentType }
       }