'instance_name' not configured for 'main' (but not using 'main')

[rejuvenile]

I've configured my Nativelink server and clients to use an instance name of "" (that is, not 'main') and yet I still receive this error:

2025-03-12T19:40:45.605763Z ERROR nativelink_worker::local_worker: Error executing action, err: Error { code: Internal, messages: ["status: Internal, message: \"'instance_name' not configured for 'main' : Failed on find_missing_blobs() command\", details: [], metadata: MetadataMap { headers: {\"content-type\": \"application/grpc\", \"date\": \"Wed, 12 Mar 2025 19:40:45 GMT\", \"content-length\": \"0\"} }", "in GrpcStore::find_missing_blobs", "On attempt 1", "Failed to run has() on slow store", "Failed to get_part in get_part_unchunked", "---", "Sender dropped before sending EOF", "During first read of buf_channel::take()", "Failed to read stream to completion in get_part_unchunked", "During start_action"] }

[rejuvenile]

Note, when I add "main" back into the configuration, I get a similar error:

  2025-03-12T20:26:31.168589Z ERROR nativelink_service::bytestream_server: error: status: Internal, message: "'instance_name' not configured for 'bytestream://<nativelink_host>:50051/main'", details: [], metadata: MetadataMap { headers: {} }
    at nativelink-service/src/bytestream_server.rs:584
    in nativelink_service::bytestream_server::read with request: ReadRequest { resource_name: "bytestream://<nativelink_host>:50051/main/blobs/blake3/92ed18594a8cb5f1b8a46be3b5d5559c32387a4218f3ea411c7e837ee4dbf2d1/590", read_offset: 0, read_limit: 0 }
    in nativelink_util::task::http_executor
    in nativelink::services::http_connection with remote_addr: 192.168.191.244:50043, socket_addr: 0.0.0.0:50051

[palfrey]

Which version was this with? Also can you provide a copy of your config?

[palfrey]

Closing due to lack of config info, so we can't really debug this one. Feel free to reopen if you've got more data for us