Skip to main content

Platform networks

Add a network to the platform catalogue, manage the RPC endpoints the platform reaches it through, and decide whether onboarding offers a choice of network.
Open it fromPlatform adminPlatform networksRole:Platform administratorConsole address: /platform-admin/networks

Platform networks holds the platform's network catalogue: every network the DALP Instance can use, and the pool of remote procedure call (RPC) endpoints the platform reaches each one through. An RPC endpoint is the address of a node that answers chain reads and accepts transactions. Every organisation on the platform shares this catalogue, so a saved change reaches each organisation that uses the network, and the platform applies it without a restart.

Use this page to put a network in the catalogue before any organisation can add it, to change a network's endpoints or settings, or to check endpoint health. An organisation adds an enabled catalogue network to itself on its own Networks page.

The page also carries the Offer network choice at onboarding switch. It decides what the next organisation onboarding offers. When it is off, a new organisation starts on the default network. When it is on, onboarding offers every enabled network. The switch takes effect at once and changes nothing for organisations that already exist. It does not turn on an organisation's own “Multi-network support” switch. Two switches with similar names says who sets each one and in which order.

The platform network catalogue with the onboarding switch, a filter, an “Add network” button and one card per network.

The onboarding network choice switch carries its effect line. Each card shows the network name, the chain name and chain ID, “Enabled” or “Disabled”, a “Default” marker on the default network, the indexing state, and the stored RPC endpoints with their health.

Steps

Open Platform networks

In the sidebar, open “Platform admin”, then Platform networks. The switch “Offer network choice at onboarding” sits at the top, with one line that states its current effect. Below it, the catalogue shows one card per network. Type in “Filter by name or chain ID” to narrow the cards.

Add a network

Select Add network to open the “Add a network” sheet, then fill in the network definition:

  • Enter a Network name. The name identifies the network everywhere and cannot change later.
  • Under Chain definition, choose “Built-in chain” and enter a “Chain alias”, such as sepolia, or choose “Custom chain” and enter a “Chain ID” and a “Chain name”. The native currency, block explorer and Multicall3 fields describe the chain.
  • Leave Enabled on if the network should be offered as soon as its endpoints are saved.

Under “RPC endpoints”, select Add endpoint for each endpoint, then fill in its fields:

  • Choose a Provider: a hosted RPC provider that takes an API key header, “Private node” for your own node behind basic authentication, or “Local node” for a node on the local network with no credentials.
  • Enter the URL and the credential the provider needs.
  • Once the pool has two or more endpoints, set each endpoint's Priority.

The platform stores credentials encrypted, and the browser does not keep them. At least one endpoint is needed.

When a pool has more than one endpoint, the indexer sends each call to one endpoint. Set a higher Priority on the endpoint you want used first. A higher “Priority” wins. Among endpoints with the same priority, health score, then the order you entered them, decides. An endpoint that trails the chain head too far, or whose health score is too low, is left out of routing while a usable endpoint remains, and a failed call moves to the next endpoint. Priority sets an order of preference. It does not split traffic by percentage.

Select Add network. The platform saves the network in three steps: it stores the definition as disabled, then saves the endpoints, then enables the network if you left “Enabled” on. A network is never enabled without endpoints. If a step after the first fails, the network stays in the catalogue as disabled, and the sheet says which part is left to save.

The “Add a network” sheet, where you define the network and add at least one RPC endpoint.

The sheet shows the “Enabled” switch, “Network name”, “Chain definition”, the chain fields, “Advanced settings”, and the “RPC endpoints” list with “Add endpoint”.

Advanced settings holds the finality fields “Supports the finalized block tag” and “Finality confirmations”, and the chain reorganisation depth field, “Reorg depth”. Leave the depth empty: the indexer then keeps the larger of the finality distance and 64 blocks. Finality gives the right values per network type, and Chain reorganisation handling explains the depth.

Edit a network or its endpoints

On the network's card, select Edit network. The edit sheet holds the same definition fields and the endpoint list, and saving writes the whole definition. Some networks come from the deployment configuration, the network list the operator sets when installing the platform. When you change the definition of such a network, the sheet shows “This creates a registry override”: your saved definition replaces the configured one until you remove it.

The chain of a network that already has stored RPC endpoints is locked, because those endpoints answer for the current chain. To move to another chain, remove the network and add it again on the chain you want.

If someone else saved the same network after you opened the sheet, the platform refuses your definition change and overwrites none of it. Endpoint changes in the same save are kept. Reload the stored definition and make the change again.

Tune gas, funding and RPC behaviour

Open Advanced settings. Every field there is optional. Leave a field unset to keep the platform default.

Chain charges no gas is on both the add sheet and the edit sheet. Left unset, the platform probes the chain once, caches the answer, and shows what it found. Set it only to overrule a wrong probe. If you set it to “Yes” on a chain that does charge gas, funding protection turns off and an unfunded transaction fails at the node.

The edit sheet adds the tuning groups, including “Gas and funding”, “Endpoint health scoring” and “RPC call limits”. In Gas and funding, an empty field inherits its value, and the placeholder shows the inherited figure with its source underneath: “Set on this network”, “Inherited from the deployment values”, “Inherited from the chain default”, “Inherited from the platform default” or “Detected on the chain”. A value you type is fixed on this network until you clear the field. Gas settings lists every key, its unit, its accepted range and its default.

Remove a network

A network that you added in the Console shows Remove network. Removing it takes it out of the catalogue and clears its saved RPC endpoints. A network from the deployment configuration shows “Remove override” only after it has been edited here. Removing the override restores the configured network and keeps its RPC endpoints. A configured network that was never edited has no remove control.

Troubleshooting

Add network is unavailable

The catalogue must be read in full before a new name can be checked for duplicates. The button stays unavailable while the catalogue loads, when stored override status cannot be read, and when the catalogue has more than 200 networks. The page then shows the notice “Showing the first 200 networks”, with the count of networks it does not list, and the filter does not cover those. Adding a network stays unavailable until the catalogue fits one page.

A network cannot be enabled

A network with no RPC endpoints cannot be enabled, because the platform could not reach it. Add an endpoint in the edit sheet, then turn on Enabled. Enabling is also refused when another enabled network already serves the same chain ID. The message names that network. Disable or remove that network first.

Token writes answer that the requested network is disabled

When every network on a chain ID is disabled, token writes on that chain are refused with 409 core.network.NETWORK_DISABLED, error DALP-12297. The API response names the network in data.networkName. Enable that network, or enable another network on the same chain ID, then send the write again.

Saving refuses an endpoint address

The platform refuses an endpoint it would never call: a URL that embeds credentials or carries a query string or fragment, or a host name that resolves to a loopback, private, carrier-grade NAT, link-local, reserved or multicast address. Link-local covers cloud metadata services. It also refuses an endpoint that answers with a different chain ID than the network's. The network definition stays saved, the pool is not written, and the error names the refused endpoint. Enter credentials in the credential fields, and use an endpoint that resolves to a public address. Only a network whose deployment values set rpc.allowPrivateUpstreams accepts private addresses, and the Console cannot change that setting. See Networks. An endpoint that does not answer at save time is still saved.

Disabling or removing a network is refused

The platform refuses a change that would leave a chain without an enabled network while an organisation has a system pending, deploying or active on that chain. The message names the chains. A second network cannot be enabled on a chain that an enabled network already serves, so keep the network enabled and change its RPC endpoints instead.

Saving refuses a pool with no endpoints

The Platform API refuses a pool replacement that lists no endpoints with 422 (PLATFORM_RPC_POOL_EMPTY) and leaves the stored pool as it was. List at least one endpoint. To remove a pool, delete it instead of saving it empty.

Reads or transactions on a network fail right after a change

Confirm the saved endpoints are reachable from the platform, not only from your browser. Indexer and Platform API reads on that network go through the pool you saved, and the previous pool is not kept. Check the health badges on the card: Excluded explains why routing leaves an endpoint out, for example how many blocks it trails the chain head.

One endpoint is unreliable but the network still works

The platform routes around an unhealthy endpoint while another usable one remains. Selection and hysteresis gives the thresholds that leave an endpoint out of routing. Lower the unreliable endpoint's Priority, or remove it once another usable endpoint is in the pool. Priority must stay a whole number of 1 or more.

A card shows Rejected

A stored network definition that the platform cannot serve stays in the list with the reason, instead of disappearing. Examples: the stored definition contains credentials, its chain ID is already used by another network, or it would create a second default network. Edit or remove the stored definition. Add credentials through the RPC endpoint fields, never in the definition.

A gas figure I saved is not the one the platform uses

Check that you saved it on the network you meant. A typed figure applies to that one network only, so a figure saved on a disabled catalogue entry for the same chain has no effect. Reopen the edit sheet: a field you set shows your value, and an empty field shows the inherited figure with its source underneath.

A funding request expires before anyone can act on it

Raise Funding pause window (seconds) on that network. The platform default of 30 minutes, which the deployment configuration can change per network, suits a deployment where someone can top up a wallet within that time. A chain funded by hand from a treasury needs a longer window, because work that waits for funds fails when the window runs out.

A network card says new-token detection is off

The network's RPC provider refuses the log query the platform uses to find tokens it has not seen before, so the platform tracks only the tokens it already knows on that network. The card shows when the platform last checked the provider, and the platform checks again at an interval the deployment sets, 10 minutes by default. When the provider answers again, the card clears and the platform scans the period when detection was off, so deposits of new tokens made in that period are detected. The Why does a network show that new-token detection is off? entry in the Ledger indexing stalled runbook explains the cause and the fix.

A card shows Stalled or Head unknown

“Head unknown” means no RPC endpoint of the enabled network answered with the chain head. Check the endpoints and their health. “Stalled” means indexing has committed no new block for a while although the chain has moved on. The card's Pause indexing switch, shown when the network has RPC endpoints, stops and resumes indexing for the network. Pausing stops new chain data for every organisation on that network until you resume it. The Ledger indexing stalled runbook covers the causes.

Was this page helpful?