<aside> ❗ This page has been shared publicly

</aside>

Motivated by forum post: https://forum.thegraph.com/t/n-1-to-n-solution-staging-version/3735

When developers deploys a new version of subgraph deployment, they can give indexers a heads up that there will soon be an upgrade of the subgraph. This allows indexers to start indexing the new subgraph deployment before there is query demand.

For the developer, they gain a better control and support for when they can utilize the new version.

For the indexers, they can seamlessly transition into the upgraded deployment.

statusEndpointMessage identifier as subgraph deployment hash, content includes the indexer’s public status API endpoint. This should already be public and verifiable

• Optionally: Developers run the radio for a fixed period after sending stagingVersionMessage and collect the status endpoint so they can actively monitor the indexing statuses from responding indexers.

stagingVersionMessage identifier must be old version subgraph deployment hash as it is known and subscribed by indexers, content includes new subgraph deployment hash and potentially the time they plan to deploy

• Optionally: Indexers ****get alerted and send offchain command to indexer management server
• Optionally: Indexers respond with a statusEndpointMessage with the new subgraph deployment topic

Github issues

• [x] Gossip operator - modularise identity validation function
  • [x] add graphQL query subgraph { owner } to check for developer identity
  • [ ] valid_indexer(sender_address) → higher order function valid_identity(valid_fn, sender_address, Option(ipfs_hash)) and add a variation valid_developer(sender_address, ipfs_hash)
• [ ] Radio operator
  • [ ] add graphQL query indexingStatuses to indexer management server url
  • [x] Indexers periodically send syncing status messages comparison blocks’ public PoI
    • [ ] Upon received a stage message that’s validated, POST offchain command to indexer management server or directly to graph node, or send alert to slack
  • [ ] Ideally developers can use API to post a staging message, but for now let’s have them periodically send staging version message for the version developer would support developers can use one-shot CLI to send a message to the network that should notify connected indexers that there’s a new version of the subgraph available for syncing
    • [x] Message includes: old and new subgraph deployment hash, subgraph id, planned usage migration time,
    • [ ] Optional: continue radio operation with subscription to new and old deployment message channel, notify developer when they start to receive new deployment public POIs or when they stop receiving the old deployment’s pPoi. The alert indicates to them that it is a good time to publish subgraph without instability
• [x] content topic: ~~sync-stage-ipfs-hash~~ subgraph-radio-ipfs-hash
  • [x] indexer sends publicPoiMessage periodically if available
  • [x] developer sends versionUpgradeMessage by manual fill-in in CLI
    • [ ] or automatic from UI