graphprotocol/graph-node v0.27.0

on GitHub

• Store writes are now carried out in parallel to the rest of the subgraph process, improving indexing performance for subgraphs with significant store interaction. Metrics & monitoring was updated for this new pipelined process;
• This adds support for apiVersion 0.0.7, which makes receipts accessible in Ethereum event handlers. Documentation link;
• This introduces some improvements to the subgraph GraphQL API, which now supports filtering on the basis of, and filtering for entities which changed from a certain block;
• Support was added for Arweave indexing. Tendermint was renamed to Cosmos in Graph Node. These integrations are still in "beta";
• Callhandler block filtering for contract calls now works as intended (this was a longstanding bug);
• Gas costing for mappings is still set at a very high default, as we continue to benchmark and refine this metric;
• A new graphman fix block command was added to easily refresh a block in the block cache, or clear the cache for a given network;
• IPFS file fetching now uses files/stat, as object was deprecated;
• Subgraphs indexing via a Firehose can now take advantage of Firehose-side filtering;
• NEAR subgraphs can now match accounts for receipt filtering via prefixes or suffixes.

Upgrade notes

• In the case of you having custom SQL, there's a new SQL migration;
• On the pipelining of the store writes, there's now a new environment variable GRAPH_STORE_WRITE_QUEUE (default value is 5), that if set to 0, the old synchronous behaviour will come in instead. The value stands for the amount of write/revert parallel operations #3177;
• There's now support for TLS connections in the PostgreSQL notification_listener #3503;
• GraphQL HTTP and WebSocket ports can now be set via environment variables #2832;
• The genesis block can be set via the GRAPH_ETHEREUM_GENESIS_BLOCK_NUMBER env var #3650;
• There's a new experimental feature to limit the number of subgraphs for a specific web3 provider. Link for documentation;
• Two new GraphQL validation environment variables were included: ENABLE_GRAPHQL_VALIDATIONS and SILENT_GRAPHQL_VALIDATIONS, which are documented here;
• A bug fix for graphman index was landed, which fixed the behavior where if one deployment was used by multiple names would result in the command not working #3416;
• Another fix landed for graphman, the bug would allow the unassign/reassign commands to make two or more nodes index the same subgraph by mistake #3478;
• Error messages of eth RPC providers should be clearer during graph-node start up #3422;
• Env var GRAPH_STORE_CONNECTION_MIN_IDLE will no longer panic, instead it will log a warning if it exceeds the pool_size #3489;
• Failed GraphQL queries now have proper timing information in the service metrics #3508;
• Non-primary shards now can be disabled through setting the pool_size to 0 #3513;
• Queries with large results now have a query_id #3514;
• It's now possible to disable the LFU Cache by setting GRAPH_QUERY_LFU_CACHE_SHARDS to 0 #3522;
• GRAPH_ACCOUNT_TABLES env var is not supported anymore #3525;
• New documentation landed on the metadata tables;
• GRAPH_GRAPHQL_MAX_OPERATIONS_PER_CONNECTION for GraphQL subscriptions now has a default of 1000 #3735