Feat: Propagate Error Through Context

[mrkaye97]

Propagating errors in upstream step runs through the context by attaching a new field to the input that's errors, which is a dictionary of key-value pairs where the keys are the step names.

Really not sure if this is implemented in a reasonable way, but it does seem to work - would love some feedback here :)

Example input from running the on failure example in Go:

{
  "input": {},
  "errors": {
    "step-one": "\"test on failure\""
  },
  "parents": {},
  "overrides": {},
  "user_data": null,
  "triggered_by": "manual"
}

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
hatchet-docs	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Jan 17, 2025 11:29pm

[mrkaye97]

Is there a way to test this with the other SDKs? I was trying to run a Python worker against my local setup but having issues:

[ERROR] 🪓 -- 2025-01-16 14:21:25,080 - Could not put workflow: <_InactiveRpcError of RPC that terminated with:
        status = StatusCode.UNAVAILABLE
        details = "failed to connect to all addresses; last error: UNKNOWN: ipv4:127.0.0.1:443: Ssl handshake failed (TSI_PROTOCOL_FAILURE): SSL_ERROR_SSL: error:10000438:SSL routines:OPENSSL_internal:TLSV1_ALERT_INTERNAL_ERROR: Invalid certificate verification context"
        debug_error_string = "UNKNOWN:Error received from peer  {created_time:"2025-01-16T14:21:25.079756-05:00", grpc_status:14, grpc_message:"failed to connect to all addresses; last error: UNKNOWN: ipv4:127.0.0.1:443: Ssl handshake failed (TSI_PROTOCOL_FAILURE): SSL_ERROR_SSL: error:10000438:SSL routines:OPENSSL_internal:TLSV1_ALERT_INTERNAL_ERROR: Invalid certificate verification context"}"
>

[reillyse]

I need a bit more context here to understand what we are trying to do.

[mrkaye97]

I need a bit more context here to understand what we are trying to do.

Ah yeah, sorry about that 😅

A couple of people have been asking for if they can access the error that was thrown upstream in their on_failure steps, presumably to do different things depending on the error. So I was trying to hook up the Python SDK to my local dev setup to see if I could e.g. run the on_failure example there and see how the error propagated through / how it'd need to be parsed, but not sure how to do that (was having issues - error up there ^^)

[abelanger5]

Is there a way to test this with the other SDKs? I was trying to run a Python worker against my local setup but having issues:

[ERROR] 🪓 -- 2025-01-16 14:21:25,080 - Could not put workflow: <_InactiveRpcError of RPC that terminated with:
        status = StatusCode.UNAVAILABLE
        details = "failed to connect to all addresses; last error: UNKNOWN: ipv4:127.0.0.1:443: Ssl handshake failed (TSI_PROTOCOL_FAILURE): SSL_ERROR_SSL: error:10000438:SSL routines:OPENSSL_internal:TLSV1_ALERT_INTERNAL_ERROR: Invalid certificate verification context"
        debug_error_string = "UNKNOWN:Error received from peer  {created_time:"2025-01-16T14:21:25.079756-05:00", grpc_status:14, grpc_message:"failed to connect to all addresses; last error: UNKNOWN: ipv4:127.0.0.1:443: Ssl handshake failed (TSI_PROTOCOL_FAILURE): SSL_ERROR_SSL: error:10000438:SSL routines:OPENSSL_internal:TLSV1_ALERT_INTERNAL_ERROR: Invalid certificate verification context"}"
>

@mrkaye97 I usually have to trust the Caddy root cert:

HATCHET_CLIENT_TLS_ROOT_CA_FILE="/Users/belanger/Library/Application Support/Caddy/pki/authorities/local/root.crt"

Also, ensure that HATCHET_CLIENT_TLS_STRATEGY is not set